From deejay1 at srem.org  Sun Feb  1 12:58:10 2009
From: deejay1 at srem.org (=?UTF-8?B?xYF1a2FzeiBKZXJuYcWb?=)
Date: Sun, 1 Feb 2009 13:58:10 +0100
Subject: [Buildbot-devel] SVN update problem
Message-ID: <77fcfd0a0902010458pe629214yfa58e836ceb3b3af@mail.gmail.com>

Dear mailing list.

I'm new to this list, so hello to all members!

I'm trying to set up a buildbot instance to build some project I'm a
bit involved in, but our subversion repository is often failing due to
excessive load I think. So my question is - is there any possibility I
could make the update step be repeated if it fails. It usually works
after a second or third try.
My current build factory setup is like that:
f = factory.BuildFactory()
f.addStep(source.SVN(mode='copy',
svnurl='http://svn.openstreetmap.org/applications/editors/merkaartor'))
f.addStep(shell.Configure(command='qmake-qt4 Merkaartor.pro GEOIMAGE=1
OSMARENDER=1; lrelease-qt4 Merkaartor.pro'))
f.addStep(shell.Compile(command='LC_ALL=C make -j2'))

Regards,
-- 
?ukasz [DeeJay1] Jerna?



From deejay1 at srem.org  Sun Feb  1 13:28:19 2009
From: deejay1 at srem.org (=?UTF-8?B?xYF1a2FzeiBKZXJuYcWb?=)
Date: Sun, 1 Feb 2009 14:28:19 +0100
Subject: [Buildbot-devel] SVN update problem
In-Reply-To: <77fcfd0a0902010458pe629214yfa58e836ceb3b3af@mail.gmail.com>
References: <77fcfd0a0902010458pe629214yfa58e836ceb3b3af@mail.gmail.com>
Message-ID: <77fcfd0a0902010528w6eae7f90wa1e21c3007f13e34@mail.gmail.com>

On Sun, Feb 1, 2009 at 1:58 PM, ?ukasz Jerna? <deejay1 at srem.org> wrote:
> Dear mailing list.
>
> I'm new to this list, so hello to all members!

There's nothing like a good first impression - did some RTFM (or RTFC
in that case, wnere C stands for "code") and found the relevant
setting...

-- 
?ukasz [DeeJay1] Jerna?



From vittoric at wanadoo.fr  Sun Feb  1 16:35:56 2009
From: vittoric at wanadoo.fr (Claude Vittoria)
Date: Sun, 01 Feb 2009 17:35:56 +0100
Subject: [Buildbot-devel] SVN update problem
In-Reply-To: <77fcfd0a0902010528w6eae7f90wa1e21c3007f13e34@mail.gmail.com>
References: <77fcfd0a0902010458pe629214yfa58e836ceb3b3af@mail.gmail.com>
	<77fcfd0a0902010528w6eae7f90wa1e21c3007f13e34@mail.gmail.com>
Message-ID: <4985CF6C.60807@wanadoo.fr>

Hi,

Sound like :
http://osdir.com/ml/python.buildbot.devel/2008-07/msg00059.html

Claude

?ukasz Jerna? a ?crit :
> On Sun, Feb 1, 2009 at 1:58 PM, ?ukasz Jerna? <deejay1 at srem.org> wrote:
>> Dear mailing list.
>>
>> I'm new to this list, so hello to all members!
> 
> There's nothing like a good first impression - did some RTFM (or RTFC
> in that case, wnere C stands for "code") and found the relevant
> setting...
> 



From macke at yar.nu  Mon Feb  2 08:20:11 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Mon, 02 Feb 2009 09:20:11 +0100
Subject: [Buildbot-devel] Manual Scheduler trigger?
In-Reply-To: <c02dbb2a0901301301n115fc127s2e8b33982f47d30b@mail.gmail.com>
References: <c02dbb2a0901301301n115fc127s2e8b33982f47d30b@mail.gmail.com>
Message-ID: <4986ACBB.2020002@yar.nu>

Ian Peters-Campbell wrote:
> Does anyone have any advice regarding setting up manual triggers for 
> Schedulers/Builders?  During setup I would really like to get the kinks 
> worked out of the system without spamming check-ins to our SVN 
> repository, so if there was a good quick way to order the Buildmaster to 
> fire off a particular Builder at will it would help.  I guess I could 
> swap in a PeriodicScheduler with a short period to test with, but if 
> there's an easy way to trigger from the command line that would be 
> preferable.

I usually use the 'Force build' button on the waterfall, unless I'm 
testing commit-hooks. For that, I have a sandbox repo/buildbot/trac 
combo that I can play with.

Cheers,
/Marcus




From bacon at mcs.anl.gov  Mon Feb  2 12:27:34 2009
From: bacon at mcs.anl.gov (Charles Bacon)
Date: Mon, 2 Feb 2009 06:27:34 -0600
Subject: [Buildbot-devel] Manual Scheduler trigger?
In-Reply-To: <4986ACBB.2020002@yar.nu>
References: <c02dbb2a0901301301n115fc127s2e8b33982f47d30b@mail.gmail.com>
	<4986ACBB.2020002@yar.nu>
Message-ID: <2EB0083B-67F8-448A-AF53-9EE4F5C98FF4@mcs.anl.gov>

On Feb 2, 2009, at 2:20 AM, Marcus Lindblom wrote:

> I usually use the 'Force build' button on the waterfall, unless I'm
> testing commit-hooks. For that, I have a sandbox repo/buildbot/trac
> combo that I can play with.

That works right up until the spammers find your buildbot waterfall  
site and start clicking it randomly.  I too would welcome a  
commandline "fire this build now please" option, or at least a  
password on the force build button.


Charles



From macke at yar.nu  Mon Feb  2 12:54:33 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Mon, 02 Feb 2009 13:54:33 +0100
Subject: [Buildbot-devel] Manual Scheduler trigger?
In-Reply-To: <2EB0083B-67F8-448A-AF53-9EE4F5C98FF4@mcs.anl.gov>
References: <c02dbb2a0901301301n115fc127s2e8b33982f47d30b@mail.gmail.com>	<4986ACBB.2020002@yar.nu>
	<2EB0083B-67F8-448A-AF53-9EE4F5C98FF4@mcs.anl.gov>
Message-ID: <4986ED09.50606@yar.nu>

Charles Bacon wrote:
> On Feb 2, 2009, at 2:20 AM, Marcus Lindblom wrote:
> 
>> I usually use the 'Force build' button on the waterfall, unless I'm
>> testing commit-hooks. For that, I have a sandbox repo/buildbot/trac
>> combo that I can play with.
> 
> That works right up until the spammers find your buildbot waterfall  
> site and start clicking it randomly.  I too would welcome a  
> commandline "fire this build now please" option, or at least a  
> password on the force build button.

True. We're using BuildBot on our LAN only, so we're safe from malicious 
users.

Cheers,
/Marcus




From Y.Itskovich at Astronautics.com  Mon Feb  2 17:52:29 2009
From: Y.Itskovich at Astronautics.com (Yury Itskovich)
Date: Mon, 2 Feb 2009 11:52:29 -0600
Subject: [Buildbot-devel] need help using "try"
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
Message-ID: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>

Changed master.cfg to say:
try_sch = Try_Userpass("testtry", ["buildbot-full"], port=8031, userpass=[("alice","pw")] )

and in options:
try_master = 'localhost:8031'

Still get the same error.  Using the --master, --username, and --passwd options from the command line fails as well, but with a less descriptive error message.  Anyone have any input at all?  Could the problem be totally unrelated to the buildmaster hostname, etc?  

Can I use the manhole thing to get more information on what's wrong?
Any suggestions would be really appreciated...




-----Original Message-----
From: Yury Itskovich
Sent: Fri 1/30/2009 10:59 AM
To: buildbot-devel at lists.sourceforge.net
Subject: need help using "try"
 
Hello,

I'm very new to using BuildBot, and thought the easiest way to start would be to use the "try" command on an existing SVN repository.  So, I am doing "buildbot try --wait" from my local checkout tree, and get this:

$ buildbot try --wait
using 'pb' connect method
job created
job has been delivered
Traceback (most recent call last):
  File "/usr/lib/python2.5/site-packages/twisted/internet/defer.py", line 285, in unpause
    self._runCallbacks()
  File "/usr/lib/python2.5/site-packages/twisted/internet/defer.py", line 328, in _runCallbacks
    self.result = callback(self.result, *args, **kw)
  File "/usr/lib/python2.5/site-packages/twisted/internet/defer.py", line 289, in _continue
    self.unpause()
  File "/usr/lib/python2.5/site-packages/twisted/internet/defer.py", line 285, in unpause
    self._runCallbacks()
--- <exception caught here> ---
  File "/usr/lib/python2.5/site-packages/twisted/internet/defer.py", line 328, in _runCallbacks
    self.result = callback(self.result, *args, **kw)
  File "/usr/lib/python2.5/site-packages/buildbot/scripts/tryclient.py", line 639, in <lambda>
    d.addCallback(lambda res: self.getStatus())
  File "/usr/lib/python2.5/site-packages/buildbot/scripts/tryclient.py", line 476, in getStatus
    host, port = master.split(":")
exceptions.AttributeError: 'NoneType' object has no attribute 'split'
Unhandled error in Deferred:
Traceback (most recent call last):
Failure: twisted.spread.pb.PBConnectionLost: [Failure instance: Traceback (failure with no frames): <class 'twisted.internet.error.ConnectionLost'>: Connection to the other side was lost in a non-clean fashion: Connection lost.
]

from twistd.log:

2009-01-30 10:22:47-0600 [-] Log opened.
2009-01-30 10:22:47-0600 [-] twistd 8.1.0 (/usr/bin/python 2.5.2) starting up
2009-01-30 10:22:47-0600 [-] reactor class: <class 'twisted.internet.selectreactor.SelectReactor'>
2009-01-30 10:22:47-0600 [-] loading configuration from /home/yury/Buildbot/master/testproj/master.cfg
2009-01-30 10:22:47-0600 [-] adding new builder buildbot-full for category None
2009-01-30 10:22:47-0600 [-] trying to load status pickle from /home/yury/Buildbot/master/testproj/full/builder
2009-01-30 10:22:47-0600 [-] added builder buildbot-full in category None
2009-01-30 10:22:47-0600 [-] twisted.spread.pb.PBServerFactory starting on 8031
2009-01-30 10:22:47-0600 [-] Starting factory <twisted.spread.pb.PBServerFactory instance at 0x87b498c>
2009-01-30 10:22:47-0600 [-] warning: no ChangeSources specified in c['change_source']
2009-01-30 10:22:47-0600 [-] twisted.spread.pb.PBServerFactory starting on 10000
2009-01-30 10:22:47-0600 [-] Starting factory <twisted.spread.pb.PBServerFactory instance at 0x87b10cc>
2009-01-30 10:22:47-0600 [-] BuildMaster listening on port tcp:10000:interface=127.0.0.1
2009-01-30 10:22:47-0600 [-] configuration update started
2009-01-30 10:22:47-0600 [-] configuration update complete
2009-01-30 10:22:47-0600 [-] maybeStartBuild <Builder 'buildbot-full' at 142284044>: [] []
2009-01-30 10:22:56-0600 [Broker,0,127.0.0.1] <Scheduler 'testtry' at 142297932> got connection from user alice
2009-01-30 10:22:56-0600 [Broker,0,127.0.0.1] user alice requesting build on builders ['buildbot-full']
2009-01-30 10:22:56-0600 [Broker,0,127.0.0.1] maybeStartBuild <Builder 'buildbot-full' at 142284044>: [<buildbot.process.base.BuildRequest instance at 0x8811ccc>] []
2009-01-30 10:22:56-0600 [Broker,0,127.0.0.1] <Builder 'buildbot-full' at 142284044>: want to start build, but we don't have a remote


Here's my options file:

try_vc = 'svn'
try_connect = "pb"
try_topdir = "/home/yury/proj_trunk"
try_builders = ["buildbot-full"]
try_master = "127.0.0.1:8031"
try_username = "alice"
try_password = "pw"
try_wait = True

(Just using the logins and passwords from the documentation examples.  Btw, the try_wait doesn't seem to do anything -- I still have to use the --wait option on the command line, otherwise it exits right away).

And here's my master.cfg (everything else is commented out):

c = BuildmasterConfig = {}

from buildbot.buildslave import BuildSlave
c['slaves'] = [BuildSlave("testslave", "pass")]
c['slavePortnum'] = "tcp:10000:interface=127.0.0.1"

from buildbot.scheduler import Try_Userpass
try_sch = Try_Userpass("testtry", ["buildbot-full"], port="tcp:8031:interface=127.0.0.1", userpass=[("alice","pw")] )
c['schedulers'] = [try_sch]

from buildbot.process import factory
from buildbot.steps.source import SVN
from buildbot.steps.shell import Compile
f1 = factory.BuildFactory()
f1.addStep(SVN(svnurl="https://1.2.3.4/svn/trunk"))
f1.addStep(Compile(command=["make", "VAR=/path"]))

b1 = {'name': "buildbot-full",
      'slavename': "testslave",
      'builddir': "full",
      'factory': f1,
      }

c['builders'] = [b1]

c['status'] = []


Any ideas on what could be wrong here?


Thanks in advance,
Yury

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090202/7a6d13ca/attachment.html>

From mailtomanish.sharma at gmail.com  Mon Feb  2 18:01:54 2009
From: mailtomanish.sharma at gmail.com (Manish)
Date: Mon, 2 Feb 2009 23:31:54 +0530
Subject: [Buildbot-devel] Help with setting up a buildbot slave
Message-ID: <e7cdbe30902021001g49750b5eif41552a4e565bfaa@mail.gmail.com>

Hello List,

I am trying to setup a buildbot slave for John Wiegley's Ledger project.
I was able to create the slave and start it successfully.  The ping from
status page also works fine.

How to tell buildbot the location of the local clone of the project git
repo?  Can someone please share a sample config file that I need to feed
buildbot as far as VC config is involved?

Thanks
-- 
Manish



From macke at yar.nu  Mon Feb  2 08:20:11 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Mon, 02 Feb 2009 09:20:11 +0100
Subject: [Buildbot-devel] Manual Scheduler trigger?
In-Reply-To: <c02dbb2a0901301301n115fc127s2e8b33982f47d30b@mail.gmail.com>
References: <c02dbb2a0901301301n115fc127s2e8b33982f47d30b@mail.gmail.com>
Message-ID: <4986ACBB.2020002@yar.nu>



Ian Peters-Campbell wrote:
> Does anyone have any advice regarding setting up manual triggers for 
> Schedulers/Builders?  During setup I would really like to get the kinks 
> worked out of the system without spamming check-ins to our SVN 
> repository, so if there was a good quick way to order the Buildmaster to 
> fire off a particular Builder at will it would help.  I guess I could 
> swap in a PeriodicScheduler with a short period to test with, but if 
> there's an easy way to trigger from the command line that would be 
> preferable.

I usually use the 'Force build' button on the waterfall, unless I'm 
testing commit-hooks. For that, I have a sandbox repo/buildbot/trac 
combo that I can play with.

Cheers,
/Marcus




From macke at yar.nu  Mon Feb  2 12:54:33 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Mon, 02 Feb 2009 13:54:33 +0100
Subject: [Buildbot-devel] Manual Scheduler trigger?
In-Reply-To: <2EB0083B-67F8-448A-AF53-9EE4F5C98FF4@mcs.anl.gov>
References: <c02dbb2a0901301301n115fc127s2e8b33982f47d30b@mail.gmail.com>	<4986ACBB.2020002@yar.nu>
	<2EB0083B-67F8-448A-AF53-9EE4F5C98FF4@mcs.anl.gov>
Message-ID: <4986ED09.50606@yar.nu>



Charles Bacon wrote:
> On Feb 2, 2009, at 2:20 AM, Marcus Lindblom wrote:
> 
>> I usually use the 'Force build' button on the waterfall, unless I'm
>> testing commit-hooks. For that, I have a sandbox repo/buildbot/trac
>> combo that I can play with.
> 
> That works right up until the spammers find your buildbot waterfall  
> site and start clicking it randomly.  I too would welcome a  
> commandline "fire this build now please" option, or at least a  
> password on the force build button.

True. We're using BuildBot on our LAN only, so we're safe from malicious 
users.

Cheers,
/Marcus




From jean-michel.beuken at uclouvain.be  Mon Feb  2 19:32:12 2009
From: jean-michel.beuken at uclouvain.be (Jean-Michel Beuken)
Date: Mon, 02 Feb 2009 20:32:12 +0100
Subject: [Buildbot-devel] need help using "buildbot sendchange" & bzr
In-Reply-To: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
Message-ID: <49874A3C.1040309@uclouvain.be>

Hello,

I'm just beginning to learn buildbot and bzr

I would like to use in the VC , a command like :

buildbot sendchange --master='buildmasterHost:11111' --username=beuken 
--revision_number=285 --branch='5.8.0-public' SomeFile

where the username, revision_number and branch  were set by a cron 
script ( it checks nightly the active branch in the bzr repositories )

the "bzr checkout"  command should be  (concatenation of username and 
branch) :

bzr checkout --revision 285 
bzr+ssh://archives.host.net/soft/5.8/beuken/5.8.0-public

but how to create the variable bzrserver (with the username (who ?) and 
branch parameters )  in the master.cfg file for the definition of builder

------------------------
bzrserver = 'bzr+ssh://archives.host.net/soft/5.8/beuken/5.8.0-public'
all_steps = [ source.Bzr(baseURL=bzrserver, mode='clobber'),
                    
shell.ShellCommand(command=["./config/scripts/makemake"],timeout=12000, 
haltOnFailure = 1),
                    shell.Configure(command=["./configure"]),
                  ]
bzr_f = factory.BuildFactory(all_steps)
------------------------

thanks in advance,

regards

jmb




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090202/f4d9b17a/attachment.html>

From jean-michel.beuken at uclouvain.be  Mon Feb  2 21:22:36 2009
From: jean-michel.beuken at uclouvain.be (Jean-Michel Beuken)
Date: Mon, 02 Feb 2009 22:22:36 +0100
Subject: [Buildbot-devel] WithProperties
In-Reply-To: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
Message-ID: <4987641C.9050209@uclouvain.be>

Hello,

I'm tried to construct the baseURL for Bzr with the branch build property

but, this sentence doesn't worked :

source.Bzr(baseURL=shell.WithProperties("bzr+ssh://archives.abinit.org/abinit/5.8/%s","branch"), 
mode='clobber')

how to use the render() method ?

thanks in advance,

regards

jmb




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090202/2b409c6a/attachment.html>

From gary.poster at canonical.com  Mon Feb  2 22:50:18 2009
From: gary.poster at canonical.com (Gary Poster)
Date: Mon, 02 Feb 2009 17:50:18 -0500
Subject: [Buildbot-devel] need help using "buildbot sendchange" & bzr
In-Reply-To: <49874A3C.1040309@uclouvain.be>
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
	<49874A3C.1040309@uclouvain.be>
Message-ID: <1233615018.7262.234.camel@gary-ubuntu>

On Mon, 2009-02-02 at 20:32 +0100, Jean-Michel Beuken wrote:
> Hello,
> 
> I'm just beginning to learn buildbot and bzr
> 
> I would like to use in the VC , a command like :
> 
> buildbot sendchange --master='buildmasterHost:11111' --username=beuken
> --revision_number=285 --branch='5.8.0-public' SomeFile
> 
> where the username, revision_number and branch  were set by a cron
> script ( it checks nightly the active branch in the bzr repositories )

Hi Jean-Michel, forgive the fly-by, but have you checked out
contrib/bzr_buildbot.py in the trunk
(http://github.com/djmitche/buildbot/tree/master)?  It has a poller that
can run within the buildbot master to look for changes.  

Alternatively, if you run your main repo, it also has a change hook that
you can install in the main bzr server to push changes, rather than
pulling them in buildbot.

If you give it a try and run into issues, let me know.

Gary





From jean-michel.beuken at uclouvain.be  Tue Feb  3 05:40:24 2009
From: jean-michel.beuken at uclouvain.be (Jean-Michel Beuken)
Date: Tue, 03 Feb 2009 06:40:24 +0100
Subject: [Buildbot-devel] need help using "buildbot sendchange" & bzr
In-Reply-To: <1233615018.7262.234.camel@gary-ubuntu>
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
	<49874A3C.1040309@uclouvain.be>
	<1233615018.7262.234.camel@gary-ubuntu>
Message-ID: <4987D8C8.4060508@uclouvain.be>

Hello Gary,

thank for your answer...

I'm already test a "old" version of bzrbuildot hook but I prefer to use 
a poller...

then I'm tried to use your poller

my config Twisted 8.2, buildbot 0.7.9 and bzr 1.11

I'm put the "bzr_buildbot.py" in the same directory as the master.cfg

in master.cfg :

--------------------------------------------
...
import bzr_buildbot
bzrpoller = BzrPoller("bzr+ssh://archives.abinit.org/abinit/5.8/")
c['change_source'] = bzrpoller

all_steps = [ source.Bzr( ? ), mode='clobber'),
            ]
bzr_f = factory.BuildFactory(all_steps)

b3 = {'name': "buildbot-chum_gcc43",
      'slavename': "amd_chum",
      'builddir': "chum",
      'factory': bzr_f,
      }
c['builders'] = [ b3]

....
--------------------------------------------


1) problem with import :

[buildmaster at abinit3 Buildbot]$ buildbot checkconfig master.cfg

Traceback (most recent call last):
  File 
"/usr/lib/python2.4/site-packages/buildbot-0.7.9-py2.4.egg/buildbot/scripts/runner.py", 
line 855, in doCheckConfig
    ConfigLoader(configFile)
  File 
"/usr/lib/python2.4/site-packages/buildbot-0.7.9-py2.4.egg/buildbot/scripts/checkconfig.py", 
line 31, in __init__
    self.loadConfig(configFile)
  File 
"/usr/lib/python2.4/site-packages/buildbot-0.7.9-py2.4.egg/buildbot/master.py", 
line 487, in loadConfig
    exec f in localDict
  File "master.cfg", line 42, in ?
    import bzr_buildbot
  File "/tmp/tmp1nAd85/bzr_buildbot.py", line 257
     changes = yield twisted.internet.threads.deferToThread(
                   ^
 SyntaxError: invalid syntax


2) must I use source.Bzr() for factory.BuildFactory() definition ? if 
Yes , must I use the same "baseURL" that the url defines in BzrPoller()

3) the organisation of  our central bzr repo  is  not  so "standard"

we have some versions ( 5.8, 5.9, 6.0 ) and in each "version" repo, each 
commiters have some subversion [private-pubic] branches

for example, for the user beuken

bzr+ssh://archives.abinit.org/abinit/5.8/beuken/5.8.0-public
bzr+ssh://archives.abinit.org/abinit/5.8/beuken/5.8.0-private
bzr+ssh://archives.abinit.org/abinit/5.8/beuken/5.8.1-public
bzr+ssh://archives.abinit.org/abinit/5.8/beuken/5.8.1-private
....
bzr+ssh://archives.abinit.org/abinit/5.9/beuken/5.9.2-public
bzr+ssh://archives.abinit.org/abinit/5.9/beuken/5.9.2-private
...
bzr+ssh://archives.abinit.org/abinit/5.9/trunk/5.9.2-public
bzr+ssh://archives.abinit.org/abinit/5.9/trunk/5.9.2-private
...

Then , I want nightly  to test only the last commited "public" branches 
in all versions and only the last subversion

for example :  if the revno has changed, buildbot will test   
'5.8/beuken/5.8.1-public' and/or  '5.9/beuken/5.9.2-public'

is it possible ?

thanks,

regards,

jmb

-- 
----------------------------------------------------
Jean-Michel Beuken
Computer Scientist
Software Engineer for ETSF

Universite catholique de Louvain-La-Neuve
Lab. PCPM/FSA, Bat. BOLTZMANN
1,Place Croix du Sud
1348 Louvain-la-Neuve BELGIUM
----------------------------------------------------
Tel : (3210) 473570 Fax : (3210) 473452
HTTP://www.pcpm.ucl.ac.be/~beuken
----------------------------------------------------



From macke at yar.nu  Tue Feb  3 08:19:10 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 03 Feb 2009 09:19:10 +0100
Subject: [Buildbot-devel] need help using "try"
In-Reply-To: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>
Message-ID: <4987FDFE.9090603@yar.nu>

Yury Itskovich wrote:
> Changed master.cfg to say:
> try_sch = Try_Userpass("testtry", ["buildbot-full"], port=8031, 
> userpass=[("alice","pw")] )
> 
> and in options:
> try_master = 'localhost:8031'
> 
> Still get the same error.  Using the --master, --username, and --passwd 
> options from the command line fails as well, but with a less descriptive 
> error message.  Anyone have any input at all?  Could the problem be 
> totally unrelated to the buildmaster hostname, etc? 
> 
> Can I use the manhole thing to get more information on what's wrong?
> Any suggestions would be really appreciated...

I don't know offhand. I know I have gotten it to work, so it's not 
totally broken. :)

What are the 'less descriptive' error messages?

Cheers,
/Marcus




From macke at yar.nu  Tue Feb  3 08:25:45 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 03 Feb 2009 09:25:45 +0100
Subject: [Buildbot-devel] Help with setting up a buildbot slave
In-Reply-To: <e7cdbe30902021001g49750b5eif41552a4e565bfaa@mail.gmail.com>
References: <e7cdbe30902021001g49750b5eif41552a4e565bfaa@mail.gmail.com>
Message-ID: <4987FF89.6070407@yar.nu>

Manish wrote:
> Hello List,
> 
> I am trying to setup a buildbot slave for John Wiegley's Ledger project.
> I was able to create the slave and start it successfully.  The ping from
> status page also works fine.
> 
> How to tell buildbot the location of the local clone of the project git
> repo?  

The Git source step creates a local clone on first run, in the builddir 
parameter you specify when confuguring a builder. The build dirs are 
located in the build slave's directory-

> Can someone please share a sample config file that I need to feed
> buildbot as far as VC config is involved?

Look at the documentation on BuildFactory (6.3.2 in 0.7.9) and modify 
accordingly:

f = factory.BuildFactory()
f.addStep(source.Git(repourl="git://myserver/myrepo", mode="update"))
f.addStep(...)

b1 = {'name': "",
       'slavename': "slave1",
       'builddir': "build",
       'factory': f,
       }
c['builders'] = [ b1 ]

Cheers,
/Marcus




From Y.Itskovich at Astronautics.com  Tue Feb  3 15:13:48 2009
From: Y.Itskovich at Astronautics.com (Yury Itskovich)
Date: Tue, 3 Feb 2009 09:13:48 -0600
Subject: [Buildbot-devel] need help using "try"
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>
	<4987FDFE.9090603@yar.nu>
Message-ID: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDD@HQMAIL1.aca.astronautics.com>

Hi Marcus,

This is the output:

$ buildbot try --master=localhost:8031 --username="alice" --passwd="pw" --wait
using 'pb' connect method
job created
job has been delivered
contacting the status port at localhost:8031
Unhandled error in Deferred:
Traceback from remote host -- Traceback unavailable
buildbot-full: no build

from twistd.log:

2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] <Scheduler 'testtry' at 142722988> got connection from user alice
2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] user alice requesting build on builders ['buildbot-full']
2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] maybeStartBuild <Builder 'buildbot-full' at 144934220>: [<buildbot.process.base.BuildRequest instance at 0x8a5f3ac>, <buildbot.process.base.BuildRequest instance at 0x8a4e64c>, <buildbot.process.base.BuildRequest instance at 0x8abc74c>, <buildbot.process.base.BuildRequest instance at 0x8ac1aec>] []
2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] <Builder 'buildbot-full' at 144934220>: want to start build, but we don't have a remote
2009-02-03 09:01:36-0600 [Broker,5,127.0.0.1] Peer will receive following PB traceback:
2009-02-03 09:01:36-0600 [Broker,5,127.0.0.1] Unhandled Error
	Traceback (most recent call last):
	Failure: twisted.cred.error.UnauthorizedLogin:

master.cfg:

from buildbot.scheduler import Scheduler
c['schedulers'] = []
c['schedulers'].append(Scheduler(name="all", branch=None,
                                 treeStableTimer=2*60,
                                 builderNames=["buildbot-full"]))

from buildbot.scheduler import Try_Userpass
c['schedulers'].append(Try_Userpass("testtry", ["buildbot-full"], port=8031, userpass=[("alice","pw")]))


I created a local SVN repository, and when I commit a change, I get the same error in the log:

2009-02-02 14:36:01-0600 [-] SVNPoller polling
2009-02-02 14:36:01-0600 [-] SVNPoller: svnurl=file:///home/yury/test_repo, root=file:///home/yury/test_repo, so prefix=
2009-02-02 14:36:01-0600 [-] svnPoller: _process_changes 2 .. 3
2009-02-02 14:36:01-0600 [-] Adding change revision 3
2009-02-02 14:36:01-0600 [-] adding change, who yury, 1 files, rev=3, branch=None, comments sdf
	
2009-02-02 14:36:01-0600 [-] <Scheduler 'all' at 142679788>: change is important, adding <buildbot.changes.changes.Change instance at 0x8a591ec>
2009-02-02 14:36:01-0600 [-] <Scheduler 'all' at 142679788>: setting timer to 14:36:11
2009-02-02 14:36:01-0600 [-] SVNPoller finished polling
2009-02-02 14:36:01-0600 [-] _finished : None
2009-02-02 14:36:11-0600 [-] maybeStartBuild <Builder 'buildbot-full' at 144937836>: [<buildbot.process.base.BuildRequest instance at 0x8a5972c>] []
2009-02-02 14:36:11-0600 [-] <Builder 'buildbot-full' at 144937836>: want to start build, but we don't have a remote

I must be missing something simple in the setup/configuration that's causing this.
This is the setup for the poller and builder from master.cfg:

from buildbot.process import factory
from buildbot.steps.source import SVN
from buildbot.steps.shell import Compile
f1 = factory.BuildFactory()

f1.addStep(SVN(svnurl="file:///home/yury/test_repo"))
f1.addStep(Compile(command=["make"]))

b1 = {'name': "buildbot-full",
      'slavename': "testslave",
      'builddir': "full",
      'factory': f1,
      }

c['builders'] = [b1]


Thanks,
Yury



-----Original Message-----
From: Marcus Lindblom [mailto:macke at yar.nu]
Sent: Tue 2/3/2009 2:19 AM
To: Yury Itskovich
Cc: public-buildbot-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f at hugh.gmane.org
Subject: Re: need help using "try"
 


Yury Itskovich wrote:
> Changed master.cfg to say:
> try_sch = Try_Userpass("testtry", ["buildbot-full"], port=8031, 
> userpass=[("alice","pw")] )
> 
> and in options:
> try_master = 'localhost:8031'
> 
> Still get the same error.  Using the --master, --username, and --passwd 
> options from the command line fails as well, but with a less descriptive 
> error message.  Anyone have any input at all?  Could the problem be 
> totally unrelated to the buildmaster hostname, etc? 
> 
> Can I use the manhole thing to get more information on what's wrong?
> Any suggestions would be really appreciated...

I don't know offhand. I know I have gotten it to work, so it's not 
totally broken. :)

What are the 'less descriptive' error messages?

Cheers,
/Marcus




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090203/f3061d58/attachment.html>

From macke at yar.nu  Tue Feb  3 08:19:10 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 03 Feb 2009 09:19:10 +0100
Subject: [Buildbot-devel] need help using "try"
In-Reply-To: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>
Message-ID: <4987FDFE.9090603@yar.nu>



Yury Itskovich wrote:
> Changed master.cfg to say:
> try_sch = Try_Userpass("testtry", ["buildbot-full"], port=8031, 
> userpass=[("alice","pw")] )
> 
> and in options:
> try_master = 'localhost:8031'
> 
> Still get the same error.  Using the --master, --username, and --passwd 
> options from the command line fails as well, but with a less descriptive 
> error message.  Anyone have any input at all?  Could the problem be 
> totally unrelated to the buildmaster hostname, etc? 
> 
> Can I use the manhole thing to get more information on what's wrong?
> Any suggestions would be really appreciated...

I don't know offhand. I know I have gotten it to work, so it's not 
totally broken. :)

What are the 'less descriptive' error messages?

Cheers,
/Marcus




From macke at yar.nu  Tue Feb  3 08:25:45 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 03 Feb 2009 09:25:45 +0100
Subject: [Buildbot-devel] Help with setting up a buildbot slave
In-Reply-To: <e7cdbe30902021001g49750b5eif41552a4e565bfaa@mail.gmail.com>
References: <e7cdbe30902021001g49750b5eif41552a4e565bfaa@mail.gmail.com>
Message-ID: <4987FF89.6070407@yar.nu>



Manish wrote:
> Hello List,
> 
> I am trying to setup a buildbot slave for John Wiegley's Ledger project.
> I was able to create the slave and start it successfully.  The ping from
> status page also works fine.
> 
> How to tell buildbot the location of the local clone of the project git
> repo?  

The Git source step creates a local clone on first run, in the builddir 
parameter you specify when confuguring a builder. The build dirs are 
located in the build slave's directory-

> Can someone please share a sample config file that I need to feed
> buildbot as far as VC config is involved?

Look at the documentation on BuildFactory (6.3.2 in 0.7.9) and modify 
accordingly:

f = factory.BuildFactory()
f.addStep(source.Git(repourl="git://myserver/myrepo", mode="update"))
f.addStep(...)

b1 = {'name': "",
       'slavename': "slave1",
       'builddir': "build",
       'factory': f,
       }
c['builders'] = [ b1 ]

Cheers,
/Marcus





From macke at yar.nu  Tue Feb  3 12:19:00 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 03 Feb 2009 13:19:00 +0100
Subject: [Buildbot-devel] Help with setting up a buildbot slave
In-Reply-To: <e7cdbe30902030122r5e57bdd6oabee5c60f897ebae@mail.gmail.com>
References: <e7cdbe30902021001g49750b5eif41552a4e565bfaa@mail.gmail.com>	
	<4987FF89.6070407@yar.nu>
	<e7cdbe30902030122r5e57bdd6oabee5c60f897ebae@mail.gmail.com>
Message-ID: <49883634.7000807@yar.nu>



Manish wrote:
> On Tue, Feb 3, 2009 at 1:55 PM, Marcus Lindblom wrote:
>   
>> Manish wrote:
>>     
>>> Can someone please share a sample config file that I need to feed
>>> buildbot as far as VC config is involved?
>>>       
>> Look at the documentation on BuildFactory (6.3.2 in 0.7.9) and modify
>> accordingly:
>>
>> f = factory.BuildFactory()
>> f.addStep(source.Git(repourl="git://myserver/myrepo", mode="update"))
>> f.addStep(...)
>>
>> b1 = {'name': "",
>>   'slavename': "slave1",
>>   'builddir': "build",
>>   'factory': f,
>>   }
>> c['builders'] = [ b1 ]
>>     
> This seems to be in the configuration for for buildbot master.  I am
> only running a slave.  But thanks for the information.  This should be
> useful in future.
>   
Only running a slave won't do you much good. You need a master too, as 
the 'brain' is located there. The slaves just do as they're told, and 
sit idle if there are no one there to boss them around.

/Marcus





From mailtomanish.sharma at gmail.com  Tue Feb  3 12:22:23 2009
From: mailtomanish.sharma at gmail.com (Manish)
Date: Tue, 3 Feb 2009 17:52:23 +0530
Subject: [Buildbot-devel] Help with setting up a buildbot slave
In-Reply-To: <49883634.7000807@yar.nu>
References: <e7cdbe30902021001g49750b5eif41552a4e565bfaa@mail.gmail.com>
	<4987FF89.6070407@yar.nu>
	<e7cdbe30902030122r5e57bdd6oabee5c60f897ebae@mail.gmail.com>
	<49883634.7000807@yar.nu>
Message-ID: <e7cdbe30902030422w1c3cd279r2f2340a8942a8eac@mail.gmail.com>



On Tue, Feb 3, 2009 at 5:49 PM, Marcus Lindblom wrote:
> Manish wrote:
[snip]
>> This seems to be in the configuration for for buildbot master. I am
>> only running a slave. But thanks for the information. This should be
>> useful in future.
>>
>
> Only running a slave won't do you much good. You need a master too, as the
> 'brain' is located there. The slaves just do as they're told, and sit idle
> if there are no one there to boss them around.

Yep, got it.  I am only running a slave for John Wiegley's Ledger
buildbot for now [1] [2]. :)

1. http://github.com/jwiegley/ledger/tree/master
2. http://newartisans.com:9090/

-- 
Manish




From Y.Itskovich at Astronautics.com  Tue Feb  3 21:00:43 2009
From: Y.Itskovich at Astronautics.com (Yury Itskovich)
Date: Tue, 3 Feb 2009 15:00:43 -0600
Subject: [Buildbot-devel] need help using "try"
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>
	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>
	<4987FDFE.9090603@yar.nu>
	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDD@HQMAIL1.aca.astronautics.com>
Message-ID: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDE@HQMAIL1.aca.astronautics.com>

I should mention, I'm using:
 
Buildbot version: 0.7.8
Twisted version: 8.1.0
Python 2.5.2


Yury


-----Original Message-----
From: Yury Itskovich
Sent: Tue 2/3/2009 9:13 AM
To: Marcus Lindblom
Cc: buildbot-devel at lists.sourceforge.net
Subject: RE: need help using "try"
 
Hi Marcus,

This is the output:

$ buildbot try --master=localhost:8031 --username="alice" --passwd="pw" --wait
using 'pb' connect method
job created
job has been delivered
contacting the status port at localhost:8031
Unhandled error in Deferred:
Traceback from remote host -- Traceback unavailable
buildbot-full: no build

from twistd.log:

2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] <Scheduler 'testtry' at 142722988> got connection from user alice
2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] user alice requesting build on builders ['buildbot-full']
2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] maybeStartBuild <Builder 'buildbot-full' at 144934220>: [<buildbot.process.base.BuildRequest instance at 0x8a5f3ac>, <buildbot.process.base.BuildRequest instance at 0x8a4e64c>, <buildbot.process.base.BuildRequest instance at 0x8abc74c>, <buildbot.process.base.BuildRequest instance at 0x8ac1aec>] []
2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] <Builder 'buildbot-full' at 144934220>: want to start build, but we don't have a remote
2009-02-03 09:01:36-0600 [Broker,5,127.0.0.1] Peer will receive following PB traceback:
2009-02-03 09:01:36-0600 [Broker,5,127.0.0.1] Unhandled Error
	Traceback (most recent call last):
	Failure: twisted.cred.error.UnauthorizedLogin:

master.cfg:

from buildbot.scheduler import Scheduler
c['schedulers'] = []
c['schedulers'].append(Scheduler(name="all", branch=None,
                                 treeStableTimer=2*60,
                                 builderNames=["buildbot-full"]))

from buildbot.scheduler import Try_Userpass
c['schedulers'].append(Try_Userpass("testtry", ["buildbot-full"], port=8031, userpass=[("alice","pw")]))


I created a local SVN repository, and when I commit a change, I get the same error in the log:

2009-02-02 14:36:01-0600 [-] SVNPoller polling
2009-02-02 14:36:01-0600 [-] SVNPoller: svnurl=file:///home/yury/test_repo, root=file:///home/yury/test_repo, so prefix=
2009-02-02 14:36:01-0600 [-] svnPoller: _process_changes 2 .. 3
2009-02-02 14:36:01-0600 [-] Adding change revision 3
2009-02-02 14:36:01-0600 [-] adding change, who yury, 1 files, rev=3, branch=None, comments sdf
	
2009-02-02 14:36:01-0600 [-] <Scheduler 'all' at 142679788>: change is important, adding <buildbot.changes.changes.Change instance at 0x8a591ec>
2009-02-02 14:36:01-0600 [-] <Scheduler 'all' at 142679788>: setting timer to 14:36:11
2009-02-02 14:36:01-0600 [-] SVNPoller finished polling
2009-02-02 14:36:01-0600 [-] _finished : None
2009-02-02 14:36:11-0600 [-] maybeStartBuild <Builder 'buildbot-full' at 144937836>: [<buildbot.process.base.BuildRequest instance at 0x8a5972c>] []
2009-02-02 14:36:11-0600 [-] <Builder 'buildbot-full' at 144937836>: want to start build, but we don't have a remote

I must be missing something simple in the setup/configuration that's causing this.
This is the setup for the poller and builder from master.cfg:

from buildbot.process import factory
from buildbot.steps.source import SVN
from buildbot.steps.shell import Compile
f1 = factory.BuildFactory()

f1.addStep(SVN(svnurl="file:///home/yury/test_repo"))
f1.addStep(Compile(command=["make"]))

b1 = {'name': "buildbot-full",
      'slavename': "testslave",
      'builddir': "full",
      'factory': f1,
      }

c['builders'] = [b1]


Thanks,
Yury



-----Original Message-----
From: Marcus Lindblom [mailto:macke at yar.nu]
Sent: Tue 2/3/2009 2:19 AM
To: Yury Itskovich
Cc: public-buildbot-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f at hugh.gmane.org
Subject: Re: need help using "try"
 


Yury Itskovich wrote:
> Changed master.cfg to say:
> try_sch = Try_Userpass("testtry", ["buildbot-full"], port=8031, 
> userpass=[("alice","pw")] )
> 
> and in options:
> try_master = 'localhost:8031'
> 
> Still get the same error.  Using the --master, --username, and --passwd 
> options from the command line fails as well, but with a less descriptive 
> error message.  Anyone have any input at all?  Could the problem be 
> totally unrelated to the buildmaster hostname, etc? 
> 
> Can I use the manhole thing to get more information on what's wrong?
> Any suggestions would be really appreciated...

I don't know offhand. I know I have gotten it to work, so it's not 
totally broken. :)

What are the 'less descriptive' error messages?

Cheers,
/Marcus





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090203/9665a184/attachment.html>

From mahatmamanic at gmail.com  Tue Feb  3 21:12:49 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Tue, 3 Feb 2009 13:12:49 -0800
Subject: [Buildbot-devel] Manual Scheduler trigger?
In-Reply-To: <374457760.3310331233350514378.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
References: <1527897489.3309521233350251387.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<374457760.3310331233350514378.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
Message-ID: <c02dbb2a0902031312i1ce64e6dxbab64bd86f101771@mail.gmail.com>

Heya,

Following your advice I've set up IRC on the buildserver (handy since the
company wanted an internal IRC network anyhow.)  I'm ready to start
instantiating some IRC Bots, but I'm not seeing an easy way to limit the
builders each IRC bot has access to.

To clarify, what I would like to do is have a separate channel for each
project/dev team here at the office, and have a bot capable of performing
all of the builds for that specific project reside in the channel.  The
channel could then be locked to ensure that only team members could fire off
builds, and the build results would alwas be listed in the appropriate
channel.

Is there an easy way through existing code to create bots with a subset of
builders that would allow this, or will I need to extend the bot
functionality to get that behavior?

Ian


On Fri, Jan 30, 2009 at 1:21 PM, Stephen Milner <smilner at redhat.com> wrote:

> Hello Ian,
>
> The way I do manual builds with our dev teams is with the IRC bots. When a
> dev lead is ready to have something built they fire it off in IRC and it
> goes through whatever build/test/package/etc.. steps that are defined.
>
> Why not have two build targets .... one for internal team building (builds
> on all check ins and/or when developers say build) and one for cutting a
> release that pushes it up to the company wide repository? That way your not
> adding in build related data to check-ins.
>
> As of right now I believe there is only an XMLRPC interface to see status,
> but not to kick off new builds and things like that ... but that sounds like
> a good addition that could be done.
>
> ----- Original Message -----
> From: "Ian Peters-Campbell" <mahatmamanic at gmail.com>
> To: Buildbot-devel at lists.sourceforge.net
> Sent: Friday, January 30, 2009 4:01:58 PM GMT -05:00 US/Canada Eastern
> Subject: [Buildbot-devel] Manual Scheduler trigger?
>
>
> Does anyone have any advice regarding setting up manual triggers for
> Schedulers/Builders? During setup I would really like to get the kinks
> worked out of the system without spamming check-ins to our SVN repository,
> so if there was a good quick way to order the Buildmaster to fire off a
> particular Builder at will it would help. I guess I could swap in a
> PeriodicScheduler with a short period to test with, but if there's an easy
> way to trigger from the command line that would be preferable.
>
> On a related note, my final goal is to have continuous integration builds
> put together after each check-in, but for those to only be published to the
> development team for the particular project. I would then like a way to
> specify a "published" build that is packaged and placed on a company-wide
> server, and that notifies a wider portion of the dev team including PM and
> QA. My "best" ideas so far were to have either an SVN commit message tag
> which tagged the build as public, or a README/release notes file that was
> watched by a scheduler for the public release. I assume that others have
> dealt with similar issues, and I am wondering if anyone has suggestions as
> to the cleanest way to manage this?
>
> Thanks
> Ian
>
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by:
> SourcForge Community
> SourceForge wants to tell your story.
> http://p.sf.net/sfu/sf-spreadtheword
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090203/82b0156c/attachment.html>

From jean-michel.beuken at uclouvain.be  Tue Feb  3 22:11:45 2009
From: jean-michel.beuken at uclouvain.be (Jean-Michel Beuken)
Date: Tue, 03 Feb 2009 23:11:45 +0100
Subject: [Buildbot-devel] ShellCommand
Message-ID: <4988C121.3040104@uclouvain.be>

Hello,

I don't understand why a lot of simple ShellCommand doesn't work in 
addStep () like


shell.ShellCommand(command=["cd","tests"])  (the dir exists in the build 
folder)

the error is :

Upon execvpe cd ['cd', 'tests'] in environment id 62318864
:Traceback (most recent call last):
  File "/usr/lib64/python2.4/site-packages/twisted/internet/process.py", 
line 396, in _fork
    executable, args, environment)
  File "/usr/lib64/python2.4/site-packages/twisted/internet/process.py", 
line 442, in _execChild
    os.execvpe(executable, args, environment)
  File "/usr/lib64/python2.4/os.py", line 350, in execvpe
    _execvpe(file, args, env)
  File "/usr/lib64/python2.4/os.py", line 379, in _execvpe
    func(fullname, *argrest)
OSError: [Errno 2] No such file or directory

but these work  for exemples :

shell.ShellCommand(command=["id"])
shell.ShellCommand(command=["/bin/rm","-rf","../build"], haltOnFailure = 1)
shell.ShellCommand(command=["./config/scripts/makemake"],timeout=12000, 
haltOnFailure = 1)
shell.ShellCommand(command=["/bin/cp","doc/config/build-examples/amd-gfortran4.3_chum.ac","chum.ac"], 
haltOnFailure = 1)


thanks

regards

jmb





From markroddy at gmail.com  Tue Feb  3 22:29:58 2009
From: markroddy at gmail.com (Mark Roddy)
Date: Tue, 3 Feb 2009 17:29:58 -0500
Subject: [Buildbot-devel] ShellCommand
In-Reply-To: <4988C121.3040104@uclouvain.be>
References: <4988C121.3040104@uclouvain.be>
Message-ID: <7930f1200902031429n444547e1jd0cd299c23c3ed44@mail.gmail.com>

I noticed you didn't supply a working directory to the shell command
step.  Can you either confirm that the 'tests' exists in the root
directory of the build or try explicitly specifying the directory that
'tests' is in?  Also note that the following gotcha that you may be
falling into, the following is invalid:
shell.ShellCommand(command=["cd","proj1"])
shell.ShellCommand(command=["cd","tests"])
shell.ShellCommand(command=["make","test"])

If 'tests' is a sub directory of 'proj1'.  The current working
directory is not preserved across steps.  Instead use:
shell.ShellCommand(command=["make","test"], workdir="proj1/tests")

-Mark



On Tue, Feb 3, 2009 at 5:11 PM, Jean-Michel Beuken
<jean-michel.beuken at uclouvain.be> wrote:
> Hello,
>
> I don't understand why a lot of simple ShellCommand doesn't work in
> addStep () like
>
>
> shell.ShellCommand(command=["cd","tests"])  (the dir exists in the build
> folder)
>
> the error is :
>
> Upon execvpe cd ['cd', 'tests'] in environment id 62318864
> :Traceback (most recent call last):
>  File "/usr/lib64/python2.4/site-packages/twisted/internet/process.py",
> line 396, in _fork
>    executable, args, environment)
>  File "/usr/lib64/python2.4/site-packages/twisted/internet/process.py",
> line 442, in _execChild
>    os.execvpe(executable, args, environment)
>  File "/usr/lib64/python2.4/os.py", line 350, in execvpe
>    _execvpe(file, args, env)
>  File "/usr/lib64/python2.4/os.py", line 379, in _execvpe
>    func(fullname, *argrest)
> OSError: [Errno 2] No such file or directory
>
> but these work  for exemples :
>
> shell.ShellCommand(command=["id"])
> shell.ShellCommand(command=["/bin/rm","-rf","../build"], haltOnFailure = 1)
> shell.ShellCommand(command=["./config/scripts/makemake"],timeout=12000,
> haltOnFailure = 1)
> shell.ShellCommand(command=["/bin/cp","doc/config/build-examples/amd-gfortran4.3_chum.ac","chum.ac"],
> haltOnFailure = 1)
>
>
> thanks
>
> regards
>
> jmb
>
>
>
> ------------------------------------------------------------------------------
> Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM)
> software. With Adobe AIR, Ajax developers can use existing skills and code to
> build responsive, highly engaging applications that combine the power of local
> resources and data with the reach of the web. Download the Adobe AIR SDK and
> Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>



From dat at datascaler.com  Wed Feb  4 01:56:41 2009
From: dat at datascaler.com (Derek Taylor)
Date: Tue, 3 Feb 2009 17:56:41 -0800
Subject: [Buildbot-devel] Setting up buildbot with Git. Failed 'git-fetch'
Message-ID: <17591fe0902031756m17e4f7b0q2f2bbdb83814add@mail.gmail.com>

I'm setting up buildbot for the first time.  Our sources are in Git, stored
on github.com.

In my master.cfg:

f1 = BuildFactory()
f1.addStep(Git(repourl=("git at github.com:<repo>.git")))
f1.addStep(Compile(command=["./build.sh"]))

I tested this from the web interface, using "Force Build"

The build starts and the log file shows the following sequence of git
operations:

git-init
git-fetch
git-reset --hard HEAD

The git-reset --hard HEAD fails with "Failed to resolve 'HEAD' as a valid
ref."

This failure occurs, because the fetch doesn't seem to pull down any files.
I typically do a 'git pull' or a 'git clone' to do that.  (I'm a fairly new
git user, so I may be missing some of the finer points here.)

To get the buildbot to work, I modified slave/commands.py to execute a 'git
pull' instead of a 'git fetch' and then it works fine.  Of course, I didn't
really want to modify the buildbot sources and I suspect there is something
else I should be doing.

Did anyone else have this problem?  How should I be setting up my
BuildFactory to get the sources correctly?

Thanks!

Derek
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090203/440580b0/attachment.html>

From dtaylor at datascaler.com  Wed Feb  4 01:50:17 2009
From: dtaylor at datascaler.com (Derek Taylor)
Date: Tue, 3 Feb 2009 17:50:17 -0800
Subject: [Buildbot-devel] Setting up buildbot with Git. Failed 'git-fetch'
Message-ID: <17591fe0902031750l3afb5fcqd759821badf8e4fb@mail.gmail.com>

I'm setting up buildbot for the first time.  Our sources are in Git, stored
on github.com.

In my master.cfg:

f1 = BuildFactory()
f1.addStep(Git(repourl=("git at github.com:<repo>.git")))
f1.addStep(Compile(command=["./build.sh"]))

I tested this from the web interface, using "Force Build"

The build starts and the log file shows the following sequence of git
operations:

git-init
git-fetch
git-reset --hard HEAD

The git-reset --hard HEAD fails with "Failed to resolve 'HEAD' as a valid
ref."

This failure occurs, because the fetch doesn't seem to pull down any files.
I typically do a 'git pull' or a 'git clone' to do that.  (I'm a fairly new
git user, so I may be missing some of the finer points here.)

To get the buildbot to work, I modified slave/commands.py to execute a 'git
pull' instead of a 'git fetch' and then it works fine.  Of course, I didn't
really want to modify the buildbot sources and I suspect there is something
else I should be doing.

Did anyone else have this problem?  How should I be setting up my
BuildFactory to get the sources correctly?

Thanks!

Derek
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090203/ce431440/attachment.html>

From macke at yar.nu  Wed Feb  4 08:05:13 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Wed, 04 Feb 2009 09:05:13 +0100
Subject: [Buildbot-devel] need help using "try"
In-Reply-To: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDE@HQMAIL1.aca.astronautics.com>
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>	<4987FDFE.9090603@yar.nu>	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDD@HQMAIL1.aca.astronautics.com>
	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDE@HQMAIL1.aca.astronautics.com>
Message-ID: <49894C39.1030309@yar.nu>

Hi Yury,

Yury Itskovich wrote:

> 2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] <Builder 'buildbot-full' 
> at 144934220>: want to start build, but we don't have a remote

[snip]

> 2009-02-02 14:36:11-0600 [-] maybeStartBuild <Builder 'buildbot-full' at 
> 144937836>: [<buildbot.process.base.BuildRequest instance at 0x8a5972c>] []
> 2009-02-02 14:36:11-0600 [-] <Builder 'buildbot-full' at 144937836>: 
> want to start build, but we don't have a remote
> 
> I must be missing something simple in the setup/configuration that's 
> causing this.

'No remote' might mean no buildslave? Is the 'testslave' up and running 
properly.

I haven't seen that error message though, so I'm guesstimating. :-/

Cheers,
/Marcus




From macke at yar.nu  Wed Feb  4 08:07:13 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Wed, 04 Feb 2009 09:07:13 +0100
Subject: [Buildbot-devel] Help with setting up a buildbot slave
In-Reply-To: <e7cdbe30902030422w1c3cd279r2f2340a8942a8eac@mail.gmail.com>
References: <e7cdbe30902021001g49750b5eif41552a4e565bfaa@mail.gmail.com>	<4987FF89.6070407@yar.nu>	<e7cdbe30902030122r5e57bdd6oabee5c60f897ebae@mail.gmail.com>	<49883634.7000807@yar.nu>
	<e7cdbe30902030422w1c3cd279r2f2340a8942a8eac@mail.gmail.com>
Message-ID: <49894CB1.1080809@yar.nu>

Manish wrote:
> 
> On Tue, Feb 3, 2009 at 5:49 PM, Marcus Lindblom wrote:
>> Manish wrote:
> [snip]
>>> This seems to be in the configuration for for buildbot master. I am
>>> only running a slave. But thanks for the information. This should be
>>> useful in future.
>>>
>> Only running a slave won't do you much good. You need a master too, as the
>> 'brain' is located there. The slaves just do as they're told, and sit idle
>> if there are no one there to boss them around.
> 
> Yep, got it.  I am only running a slave for John Wiegley's Ledger
> buildbot for now [1] [2]. :)
> 
> 1. http://github.com/jwiegley/ledger/tree/master
> 2. http://newartisans.com:9090/

Well then. If John has added your slave to his buildbot config, and it 
shows up as connected on his buildbot site, you should be all set. Your 
slave will do as it's told by the master, including creatnig a clone, etc.

I get the feeling I missed the problem here? :)

Cheers,
/Marcus





From l10n.moz at googlemail.com  Wed Feb  4 10:47:21 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Wed, 4 Feb 2009 11:47:21 +0100
Subject: [Buildbot-devel] Setting up buildbot with Git. Failed
	'git-fetch'
In-Reply-To: <17591fe0902031756m17e4f7b0q2f2bbdb83814add@mail.gmail.com>
References: <17591fe0902031756m17e4f7b0q2f2bbdb83814add@mail.gmail.com>
Message-ID: <9de1bdb60902040247v4683b56dyd0e446f96682b7c1@mail.gmail.com>

It seems to me from looking at the srouce that it should actually reset to
FETCH_HEAD in 0.7.7 and beyond,
http://github.com/djmitche/buildbot/commit/bc53d760b75d89299c879e08f65c4fd99ad5506e#diff-2.
Which works perfectly fine for me locally.

The only exception would be if the given revision in your build was
explicitly HEAD, though I don't know how that would happen.

Axel

2009/2/4 Derek Taylor <dat at datascaler.com>

> I'm setting up buildbot for the first time.  Our sources are in Git, stored
> on github.com.
>
> In my master.cfg:
>
> f1 = BuildFactory()
> f1.addStep(Git(repourl=("git at github.com:<repo>.git")))
> f1.addStep(Compile(command=["./build.sh"]))
>
> I tested this from the web interface, using "Force Build"
>
> The build starts and the log file shows the following sequence of git
> operations:
>
> git-init
> git-fetch
> git-reset --hard HEAD
>
> The git-reset --hard HEAD fails with "Failed to resolve 'HEAD' as a valid
> ref."
>
> This failure occurs, because the fetch doesn't seem to pull down any
> files.  I typically do a 'git pull' or a 'git clone' to do that.  (I'm a
> fairly new git user, so I may be missing some of the finer points here.)
>
> To get the buildbot to work, I modified slave/commands.py to execute a 'git
> pull' instead of a 'git fetch' and then it works fine.  Of course, I didn't
> really want to modify the buildbot sources and I suspect there is something
> else I should be doing.
>
> Did anyone else have this problem?  How should I be setting up my
> BuildFactory to get the sources correctly?
>
> Thanks!
>
> Derek
>
> ------------------------------------------------------------------------------
> Create and Deploy Rich Internet Apps outside the browser with
> Adobe(R)AIR(TM)
> software. With Adobe AIR, Ajax developers can use existing skills and code
> to
> build responsive, highly engaging applications that combine the power of
> local
> resources and data with the reach of the web. Download the Adobe AIR SDK
> and
> Ajax docs to start building applications today-
> http://p.sf.net/sfu/adobe-com
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090204/4c567342/attachment.html>

From smilner at redhat.com  Wed Feb  4 15:46:45 2009
From: smilner at redhat.com (Stephen Milner)
Date: Wed, 4 Feb 2009 10:46:45 -0500 (EST)
Subject: [Buildbot-devel] Manual Scheduler trigger?
In-Reply-To: <963040513.4198771233761032502.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
Message-ID: <554774003.4213751233762405215.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>

Hey Ian,

The way I had set this up was having 1 master (with 1 irc  bot) to 1 group. So ...

Master A
- SlaveBuild A1 \
- SlaveBuild A2 |--------> #chanA
- SlaveBuild A3 /

Master B
- SlaveBuild B1 \
- SlaveBuild B2 |--------> #chanB

and so on..

I'm sure there are other ways to make this setup, but that is how I did it :-).


----- Original Message -----
From: "Ian Peters-Campbell" <mahatmamanic at gmail.com>
To: "Stephen Milner" <smilner at redhat.com>
Cc: Buildbot-devel at lists.sourceforge.net
Sent: Tuesday, February 3, 2009 4:12:49 PM GMT -05:00 US/Canada Eastern
Subject: Re: [Buildbot-devel] Manual Scheduler trigger?

Heya, 

Following your advice I've set up IRC on the buildserver (handy since the company wanted an internal IRC network anyhow.) I'm ready to start instantiating some IRC Bots, but I'm not seeing an easy way to limit the builders each IRC bot has access to. 

To clarify, what I would like to do is have a separate channel for each project/dev team here at the office, and have a bot capable of performing all of the builds for that specific project reside in the channel. The channel could then be locked to ensure that only team members could fire off builds, and the build results would alwas be listed in the appropriate channel. 

Is there an easy way through existing code to create bots with a subset of builders that would allow this, or will I need to extend the bot functionality to get that behavior? 

Ian 



On Fri, Jan 30, 2009 at 1:21 PM, Stephen Milner < smilner at redhat.com > wrote: 


Hello Ian, 

The way I do manual builds with our dev teams is with the IRC bots. When a dev lead is ready to have something built they fire it off in IRC and it goes through whatever build/test/package/etc.. steps that are defined. 

Why not have two build targets .... one for internal team building (builds on all check ins and/or when developers say build) and one for cutting a release that pushes it up to the company wide repository? That way your not adding in build related data to check-ins. 

As of right now I believe there is only an XMLRPC interface to see status, but not to kick off new builds and things like that ... but that sounds like a good addition that could be done. 




----- Original Message ----- 
From: "Ian Peters-Campbell" < mahatmamanic at gmail.com > 
To: Buildbot-devel at lists.sourceforge.net 
Sent: Friday, January 30, 2009 4:01:58 PM GMT -05:00 US/Canada Eastern 
Subject: [Buildbot-devel] Manual Scheduler trigger? 


Does anyone have any advice regarding setting up manual triggers for Schedulers/Builders? During setup I would really like to get the kinks worked out of the system without spamming check-ins to our SVN repository, so if there was a good quick way to order the Buildmaster to fire off a particular Builder at will it would help. I guess I could swap in a PeriodicScheduler with a short period to test with, but if there's an easy way to trigger from the command line that would be preferable. 

On a related note, my final goal is to have continuous integration builds put together after each check-in, but for those to only be published to the development team for the particular project. I would then like a way to specify a "published" build that is packaged and placed on a company-wide server, and that notifies a wider portion of the dev team including PM and QA. My "best" ideas so far were to have either an SVN commit message tag which tagged the build as public, or a README/release notes file that was watched by a scheduler for the public release. I assume that others have dealt with similar issues, and I am wondering if anyone has suggestions as to the cleanest way to manage this? 

Thanks 
Ian 


------------------------------------------------------------------------------ 
This SF.net email is sponsored by: 
SourcForge Community 
SourceForge wants to tell your story. 
http://p.sf.net/sfu/sf-spreadtheword 
_______________________________________________ 
Buildbot-devel mailing list 
Buildbot-devel at lists.sourceforge.net 
https://lists.sourceforge.net/lists/listinfo/buildbot-devel 




From felix at compsoc.nuigalway.ie  Wed Feb  4 17:14:18 2009
From: felix at compsoc.nuigalway.ie (Darragh Bailey)
Date: Wed, 4 Feb 2009 17:14:18 +0000
Subject: [Buildbot-devel] Manual Scheduler trigger?
In-Reply-To: <554774003.4213751233762405215.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
References: <963040513.4198771233761032502.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<554774003.4213751233762405215.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
Message-ID: <20090204171418.GC25541@compsoc.nuigalway.ie>

On Wed, Feb 04, 2009 at 10:46:45AM -0500, Stephen Milner wrote:
> Hey Ian,
> 
> The way I had set this up was having 1 master (with 1 irc  bot) to 1 group. So ...
> 
> Master A
> - SlaveBuild A1 \
> - SlaveBuild A2 |--------> #chanA
> - SlaveBuild A3 /
> 
> Master B
> - SlaveBuild B1 \
> - SlaveBuild B2 |--------> #chanB
> 
> and so on..
> 
> I'm sure there are other ways to make this setup, but that is how I did it :-).
> 
> 
> ----- Original Message -----
> From: "Ian Peters-Campbell" <mahatmamanic at gmail.com>
> To: "Stephen Milner" <smilner at redhat.com>
> Cc: Buildbot-devel at lists.sourceforge.net
> Sent: Tuesday, February 3, 2009 4:12:49 PM GMT -05:00 US/Canada Eastern
> Subject: Re: [Buildbot-devel] Manual Scheduler trigger?
> 
> Heya, 
> 
> Following your advice I've set up IRC on the buildserver (handy since the company wanted an internal IRC network anyhow.) I'm ready to start instantiating some IRC Bots, but I'm not seeing an easy way to limit the builders each IRC bot has access to. 
> 
> To clarify, what I would like to do is have a separate channel for each project/dev team here at the office, and have a bot capable of performing all of the builds for that specific project reside in the channel. The channel could then be locked to ensure that only team members could fire off builds, and the build results would alwas be listed in the appropriate channel. 
> 
> Is there an easy way through existing code to create bots with a subset of builders that would allow this, or will I need to extend the bot functionality to get that behavior? 
> 
> Ian 
> 

Use of categories should take care of this. Group your builds into
categories and create different IRC bot for each of the categories,
either on the same channel or different channels.


-- 
Darragh

"Nothing is foolproof to a sufficiently talented fool."



From macke at yar.nu  Wed Feb  4 08:05:13 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Wed, 04 Feb 2009 09:05:13 +0100
Subject: [Buildbot-devel] need help using "try"
In-Reply-To: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDE@HQMAIL1.aca.astronautics.com>
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>	<4987FDFE.9090603@yar.nu>	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDD@HQMAIL1.aca.astronautics.com>
	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDE@HQMAIL1.aca.astronautics.com>
Message-ID: <49894C39.1030309@yar.nu>



Hi Yury,

Yury Itskovich wrote:

> 2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] <Builder 'buildbot-full' 
> at 144934220>: want to start build, but we don't have a remote

[snip]

> 2009-02-02 14:36:11-0600 [-] maybeStartBuild <Builder 'buildbot-full' at 
> 144937836>: [<buildbot.process.base.BuildRequest instance at 0x8a5972c>] []
> 2009-02-02 14:36:11-0600 [-] <Builder 'buildbot-full' at 144937836>: 
> want to start build, but we don't have a remote
> 
> I must be missing something simple in the setup/configuration that's 
> causing this.

'No remote' might mean no buildslave? Is the 'testslave' up and running 
properly.

I haven't seen that error message though, so I'm guesstimating. :-/

Cheers,
/Marcus




From dtaylor at datascaler.com  Wed Feb  4 18:37:03 2009
From: dtaylor at datascaler.com (Derek Taylor)
Date: Wed, 4 Feb 2009 10:37:03 -0800
Subject: [Buildbot-devel] Setting up buildbot with Git. Failed
	'git-fetch'
In-Reply-To: <9de1bdb60902040247v4683b56dyd0e446f96682b7c1@mail.gmail.com>
References: <17591fe0902031756m17e4f7b0q2f2bbdb83814add@mail.gmail.com>
	<9de1bdb60902040247v4683b56dyd0e446f96682b7c1@mail.gmail.com>
Message-ID: <17591fe0902041037l5927a472p1aa09c2f057e2fd8@mail.gmail.com>

Axel,

Thanks for the reply.

My problem was in thinking it should use 'HEAD'.  I was passing this in
through the "Force Build" web page.  When I leave the field blank, it works
as you suggest.

Derek

On Wed, Feb 4, 2009 at 2:47 AM, Axel Hecht <l10n.moz at googlemail.com> wrote:

> It seems to me from looking at the srouce that it should actually reset to
> FETCH_HEAD in 0.7.7 and beyond,
> http://github.com/djmitche/buildbot/commit/bc53d760b75d89299c879e08f65c4fd99ad5506e#diff-2.
> Which works perfectly fine for me locally.
>
> The only exception would be if the given revision in your build was
> explicitly HEAD, though I don't know how that would happen.
>
> Axel
>
> 2009/2/4 Derek Taylor <dat at datascaler.com>
>
>> I'm setting up buildbot for the first time.  Our sources are in Git,
>> stored on github.com.
>>
>> In my master.cfg:
>>
>> f1 = BuildFactory()
>> f1.addStep(Git(repourl=("git at github.com:<repo>.git")))
>> f1.addStep(Compile(command=["./build.sh"]))
>>
>> I tested this from the web interface, using "Force Build"
>>
>> The build starts and the log file shows the following sequence of git
>> operations:
>>
>> git-init
>> git-fetch
>> git-reset --hard HEAD
>>
>> The git-reset --hard HEAD fails with "Failed to resolve 'HEAD' as a valid
>> ref."
>>
>> This failure occurs, because the fetch doesn't seem to pull down any
>> files.  I typically do a 'git pull' or a 'git clone' to do that.  (I'm a
>> fairly new git user, so I may be missing some of the finer points here.)
>>
>> To get the buildbot to work, I modified slave/commands.py to execute a
>> 'git pull' instead of a 'git fetch' and then it works fine.  Of course, I
>> didn't really want to modify the buildbot sources and I suspect there is
>> something else I should be doing.
>>
>> Did anyone else have this problem?  How should I be setting up my
>> BuildFactory to get the sources correctly?
>>
>> Thanks!
>>
>> Derek
>>
>> ------------------------------------------------------------------------------
>> Create and Deploy Rich Internet Apps outside the browser with
>> Adobe(R)AIR(TM)
>> software. With Adobe AIR, Ajax developers can use existing skills and code
>> to
>> build responsive, highly engaging applications that combine the power of
>> local
>> resources and data with the reach of the web. Download the Adobe AIR SDK
>> and
>> Ajax docs to start building applications today-
>> http://p.sf.net/sfu/adobe-com
>> _______________________________________________
>> Buildbot-devel mailing list
>> Buildbot-devel at lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>>
>>
>


-- 
Derek Taylor
Director of Engineering
DataScaler Inc.

201A Ravendale Drive
Mountain View, CA  94043
650-515-3447
408-702-5442 (m)
dtaylor at datascaler.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090204/ff0a10cf/attachment.html>

From mahatmamanic at gmail.com  Thu Feb  5 02:23:37 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Wed, 4 Feb 2009 18:23:37 -0800
Subject: [Buildbot-devel] debug client import error
Message-ID: <c02dbb2a0902041823s5d47321u684b1a6019f50a38@mail.gmail.com>

Hey all,

This may be beyond the purview of this list, but I am hoping someone else
ran into the same problem.  When I try to run buildbot debugclient I get an
ImportError from Zope attempting to import gobject.  When I try to install
pygobject through yum it tells me that it is already up to date (which I
hope is correct because when I try to install it manually I run into a rat's
nest of dependencies.)  I am thinking this might be as simple as a
pythonpath configuration issue or setting up a soft link somewhere.  Can
anyone point me in the right direction?

Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090204/8af2bec7/attachment.html>

From mahatmamanic at gmail.com  Thu Feb  5 21:32:33 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Thu, 5 Feb 2009 13:32:33 -0800
Subject: [Buildbot-devel] debug client import error
In-Reply-To: <c02dbb2a0902041823s5d47321u684b1a6019f50a38@mail.gmail.com>
References: <c02dbb2a0902041823s5d47321u684b1a6019f50a38@mail.gmail.com>
Message-ID: <c02dbb2a0902051332h7550552s4c6623e58db41ab3@mail.gmail.com>

Well, I have gone down quite the rabbit hole with this one.  I finally
managed to get pygbject and pygtk working with Python 2.5 but I am now
running into problems importing pango and glade when I attempt to run the
debug client.  I'm pretty frustrated with it.

Is there any other way to get useful error information out of Buildbot?
This all stemmed from getting a context-less SVN Exception when doing a
check-out, and wanting more information.  The debugclient looked perfect but
after chasing ghosts for almost a full day I would love another solution for
logging/verbose output if there is one?

Ian


On Wed, Feb 4, 2009 at 6:23 PM, Ian Peters-Campbell
<mahatmamanic at gmail.com>wrote:

> Hey all,
>
> This may be beyond the purview of this list, but I am hoping someone else
> ran into the same problem.  When I try to run buildbot debugclient I get an
> ImportError from Zope attempting to import gobject.  When I try to install
> pygobject through yum it tells me that it is already up to date (which I
> hope is correct because when I try to install it manually I run into a rat's
> nest of dependencies.)  I am thinking this might be as simple as a
> pythonpath configuration issue or setting up a soft link somewhere.  Can
> anyone point me in the right direction?
>
> Ian
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090205/33ee993b/attachment.html>

From mue.r at gmx.de  Fri Feb  6 07:54:57 2009
From: mue.r at gmx.de (=?UTF-8?B?UmVuw6kgTcO8bGxlcg==?=)
Date: Fri, 06 Feb 2009 08:54:57 +0100
Subject: [Buildbot-devel] debug client import error
In-Reply-To: <c02dbb2a0902051332h7550552s4c6623e58db41ab3@mail.gmail.com>
References: <c02dbb2a0902041823s5d47321u684b1a6019f50a38@mail.gmail.com>
	<c02dbb2a0902051332h7550552s4c6623e58db41ab3@mail.gmail.com>
Message-ID: <gmgqci$ho7$1@ger.gmane.org>

Ian Peters-Campbell schrieb:
> Well, I have gone down quite the rabbit hole with this one.  I finally
> managed to get pygbject and pygtk working with Python 2.5 but I am now
> running into problems importing pango and glade when I attempt to run
> the debug client.  I'm pretty frustrated with it. 
> 

Hi Ian,

I had some problems to get the debugclient and the statusclient running
on windows, too. I finally got a working configuration, but only with
very specific versions, I share them with you, maybe it helps:

- Python 2.5.1
- Twisted_NoDocs-2.5.0.win32-py2.5.exe
- buildbot 0.7.9
- pycairo-1.4.12-1.win32-py2.5.exe
- pygobject-2.14.1-1.win32-py2.5.exe
- pygtk-2.12.1-2.win32-py2.5.exe
- gtk-2.12.9-win32-2.exe

I hope this helps.

Greetings,
Ren?




From mahatmamanic at gmail.com  Fri Feb  6 19:43:03 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Fri, 6 Feb 2009 11:43:03 -0800
Subject: [Buildbot-devel] debug client import error
In-Reply-To: <gmgqci$ho7$1@ger.gmane.org>
References: <c02dbb2a0902041823s5d47321u684b1a6019f50a38@mail.gmail.com>
	<c02dbb2a0902051332h7550552s4c6623e58db41ab3@mail.gmail.com>
	<gmgqci$ho7$1@ger.gmane.org>
Message-ID: <c02dbb2a0902061143i6d6500f9ga521593d33d9d04f@mail.gmail.com>

Rene,

Thanks for the info, I appreciate the feedback.  With the problems I kept
running into I eventually switched my default interpreter back to Python 2.4
and reinstalled Buildbot and its dependencies under /usr rather than
/usr/local.  Rebuilding a handful of relatively stable packages turned out
to be way, way easier than basically attempting to recreate the entire glib
dependence tree (not to mention the Crypto stuff for the manhole
functionality) in a new location.

I'm guessing some pretty smart people have spent a fair chunk of time
setting up the CentOS/Gnome distro with what it needs to run, and as long as
Buildbot is stable on Python 2.4 it seems to make more sense to leverage the
work they have done than to try to recreate the entire library from scratch.

Thanks to everyone for the feedback, and if anyone has these problems in the
future I highly recommend just going with your distro's pre-installed Python
version if possible.

Ian

On Thu, Feb 5, 2009 at 11:54 PM, Ren? M?ller <mue.r at gmx.de> wrote:

> Ian Peters-Campbell schrieb:
> > Well, I have gone down quite the rabbit hole with this one.  I finally
> > managed to get pygbject and pygtk working with Python 2.5 but I am now
> > running into problems importing pango and glade when I attempt to run
> > the debug client.  I'm pretty frustrated with it.
> >
>
> Hi Ian,
>
> I had some problems to get the debugclient and the statusclient running
> on windows, too. I finally got a working configuration, but only with
> very specific versions, I share them with you, maybe it helps:
>
> - Python 2.5.1
> - Twisted_NoDocs-2.5.0.win32-py2.5.exe
> - buildbot 0.7.9
> - pycairo-1.4.12-1.win32-py2.5.exe
> - pygobject-2.14.1-1.win32-py2.5.exe
> - pygtk-2.12.1-2.win32-py2.5.exe
> - gtk-2.12.9-win32-2.exe
>
> I hope this helps.
>
> Greetings,
> Ren?
>
>
>
> ------------------------------------------------------------------------------
> Create and Deploy Rich Internet Apps outside the browser with
> Adobe(R)AIR(TM)
> software. With Adobe AIR, Ajax developers can use existing skills and code
> to
> build responsive, highly engaging applications that combine the power of
> local
> resources and data with the reach of the web. Download the Adobe AIR SDK
> and
> Ajax docs to start building applications today-
> http://p.sf.net/sfu/adobe-com
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090206/d1715b5b/attachment.html>

From mue.r at gmx.de  Mon Feb  9 20:49:09 2009
From: mue.r at gmx.de (=?ISO-8859-15?Q?Ren=E9_M=FCller?=)
Date: Mon, 09 Feb 2009 21:49:09 +0100
Subject: [Buildbot-devel] DirectoryUpload
In-Reply-To: <493597A2.30503@yar.nu>
References: <ggm65p$uo6$1@ger.gmane.org>	<492ED4AF.7080202@yar.nu>	<ggo8au$bht$1@ger.gmane.org>	<492FA7FA.5060307@yar.nu>	<gh3g6v$kv5$1@ger.gmane.org>
	<493597A2.30503@yar.nu>
Message-ID: <gmq4s5$oul$1@ger.gmane.org>

Marcus Lindblom schrieb:
> Ren? M?ller wrote:
>> Marcus Lindblom schrieb:
>>> If not, then I guess the DirectoryUpload is a workable solution. I think 
>>> it's useful anyway, as it simplifies things a bit.
>> I've made a buildstep and opened a ticket for it (#393). Please take a
>> look at it. It is the first time I worked with twisted, maybe the code
>> will look ugly.
> 
> I'll leave it to Dustin and Brian to comment on the quality. (I'm just 
> learning to hack buildbot myself, anyway, it looks pretty tidy to me.)
> 
> I do know that they usually require some documentation and tests in 
> order to merge patches into the repo, though, so you might want to work 
> on that.
> 

I completely remade the patch and now it works great. I added some
tests, too. What documentation is needed to get a review?

Ren?




From mahatmamanic at gmail.com  Tue Feb 10 01:36:23 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Mon, 9 Feb 2009 17:36:23 -0800
Subject: [Buildbot-devel] Overriding default IRC behavior
Message-ID: <c02dbb2a0902091736m212af484k1fbf6b187679ea2f@mail.gmail.com>

For the purposes of our build system I'd like for users to be able to pass
some build properties along with the build request they send to the IRC
bot.  Looking at the code in words.py it looks like the easiest way to do
that would be to

1) extend IRC, IRCStatusFactory, IRCStatusBot, and IRCContact to override
some of the functionality
2) have keyword style arguments stripped from the data coming into
IRCStatusBot.action
3) passing them down the stack in a dict to IRCContact.command_FORCE
4) finally passing them into the BuildRequest constructor.

Has anyone done anything similar and does that seem like a reasonable
approach, or is there a pre-existing trick or a way to write less code that
will let my teammates specify things like target servers or signing
credentials when requesting a build?

Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090209/98ed90ff/attachment.html>

From stephen at starswan.com  Tue Feb 10 12:17:22 2009
From: stephen at starswan.com (Stephen Dicks)
Date: Tue, 10 Feb 2009 12:17:22 +0000
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
Message-ID: <49917052.80706@starswan.com>

Hi,
    Having introduced buildbot to several companies over the last few
years, right now I find myself in what some contractors call 'on the
bench' (ie looking for work). Buildbot seems to have made quite slow
progress recently (I started with 0.7.4 in 2005), so I would like to
give it a 'shot in the arm' if I can - but I would like some advice as
to which areas I should tackle. I can see that there are a large number
of trac tickets outstanding and also the 'shimmering vision' of BuildBot
1.0 - but I'm not sure where to start.

I think the meta-buildbot needs resurrecting also - its building code 4
months out of date. How can I get access to its configuration? Should it
be switched over to the git repositories, or should the darcs one be
updated somehow? (It seems that the win32 git client is still not really
stable, which is an issue - and the win32 slave appears to be
permanently offline). I'm currently scratching my own itch with BB which
is its lack of a cross-platform monitoring GUI application, but would
like to attack the main codebase once I'm done.

All pointers gratefully received.

--
Steve Dicks



From smilner at redhat.com  Tue Feb 10 15:48:23 2009
From: smilner at redhat.com (Stephen Milner)
Date: Tue, 10 Feb 2009 10:48:23 -0500 (EST)
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
Message-ID: <1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>

Note: This is my $0.02 ... nothing more.

I think attacking tickets and your own itches are the best way at it ... at least until there is a major cut over to 1.0 for development as there are 0.x updates that are probably merging into the 1.0 codebase. More than likely if your hitting something, so are other people :-).

----- Original Message -----
From: "Stephen Dicks" <stephen at starswan.com>
To: buildbot-devel at lists.sourceforge.net
Sent: Tuesday, February 10, 2009 7:17:22 AM GMT -05:00 US/Canada Eastern
Subject: [Buildbot-devel] Full-Time buildbot developer available...!

Hi,
    Having introduced buildbot to several companies over the last few
years, right now I find myself in what some contractors call 'on the
bench' (ie looking for work). Buildbot seems to have made quite slow
progress recently (I started with 0.7.4 in 2005), so I would like to
give it a 'shot in the arm' if I can - but I would like some advice as
to which areas I should tackle. I can see that there are a large number
of trac tickets outstanding and also the 'shimmering vision' of BuildBot
1.0 - but I'm not sure where to start.

I think the meta-buildbot needs resurrecting also - its building code 4
months out of date. How can I get access to its configuration? Should it
be switched over to the git repositories, or should the darcs one be
updated somehow? (It seems that the win32 git client is still not really
stable, which is an issue - and the win32 slave appears to be
permanently offline). I'm currently scratching my own itch with BB which
is its lack of a cross-platform monitoring GUI application, but would
like to attack the main codebase once I'm done.

All pointers gratefully received.

--
Steve Dicks

------------------------------------------------------------------------------
Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM)
software. With Adobe AIR, Ajax developers can use existing skills and code to
build responsive, highly engaging applications that combine the power of local
resources and data with the reach of the web. Download the Adobe AIR SDK and
Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com
_______________________________________________
Buildbot-devel mailing list
Buildbot-devel at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/buildbot-devel



From deejay1 at srem.org  Tue Feb 10 15:56:38 2009
From: deejay1 at srem.org (=?UTF-8?B?xYF1a2FzeiBKZXJuYcWb?=)
Date: Tue, 10 Feb 2009 16:56:38 +0100
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
Message-ID: <77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>

Hi!

Please please please add usable multiple project support with one
buildmaster. Maintaining 20 different deamons is depressing ;)

But it's just mine 0.02 PLN of course ;)

Regards,
-- 
?ukasz [DeeJay1] Jerna?



From smilner at redhat.com  Tue Feb 10 16:02:50 2009
From: smilner at redhat.com (Stephen Milner)
Date: Tue, 10 Feb 2009 11:02:50 -0500 (EST)
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
Message-ID: <612402910.5545701234281770551.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>

That is a good one to add!

In the meantime a co-worker of mine wrote a buildbot instance manager called Loki (https://fedorahosted.org/loki/) which might prove useful for managing multiple daemons across multiple build nodes. It sits on top of Func and Buildbot, uses Director for CLI and a few other nice things.

Example Usage: https://fedorahosted.org/loki/wiki/ManagingBuildBots


----- Original Message -----
From: "?ukasz Jerna?" <deejay1 at srem.org>
To: buildbot-devel at lists.sourceforge.net
Sent: Tuesday, February 10, 2009 10:56:38 AM GMT -05:00 US/Canada Eastern
Subject: Re: [Buildbot-devel] Full-Time buildbot developer available...!

Hi!

Please please please add usable multiple project support with one
buildmaster. Maintaining 20 different deamons is depressing ;)

But it's just mine 0.02 PLN of course ;)

Regards,
-- 
?ukasz [DeeJay1] Jerna?

------------------------------------------------------------------------------
Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM)
software. With Adobe AIR, Ajax developers can use existing skills and code to
build responsive, highly engaging applications that combine the power of local
resources and data with the reach of the web. Download the Adobe AIR SDK and
Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com
_______________________________________________
Buildbot-devel mailing list
Buildbot-devel at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/buildbot-devel



From zandr at allmydata.com  Tue Feb 10 16:05:57 2009
From: zandr at allmydata.com (Aleksandr Milewski)
Date: Tue, 10 Feb 2009 08:05:57 -0800
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
Message-ID: <03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>

Agreed, this is very high on my list.

I'd also like to be able to subscribe a single buildslave to multiple  
buildmasters.

Not sure if there's a ticket on this, I've been spoiled by being able  
to talk to warner across the room. :)

-Zandr


On Feb 10, 2009, at 7:56 AM, ?ukasz Jerna? wrote:

> Hi!
>
> Please please please add usable multiple project support with one
> buildmaster. Maintaining 20 different deamons is depressing ;)
>
> But it's just mine 0.02 PLN of course ;)
>
> Regards,
> -- 
> ?ukasz [DeeJay1] Jerna?
>
> ------------------------------------------------------------------------------
> Create and Deploy Rich Internet Apps outside the browser with  
> Adobe(R)AIR(TM)
> software. With Adobe AIR, Ajax developers can use existing skills  
> and code to
> build responsive, highly engaging applications that combine the  
> power of local
> resources and data with the reach of the web. Download the Adobe AIR  
> SDK and
> Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel




From zandr at allmydata.com  Tue Feb 10 16:14:47 2009
From: zandr at allmydata.com (Aleksandr Milewski)
Date: Tue, 10 Feb 2009 08:14:47 -0800
Subject: [Buildbot-devel] Disposable VM Slaves?
Message-ID: <1F7FCC76-0499-4931-A201-C3E8F944FCEB@allmydata.com>

Does anyone have a good solution for running buildslaves on disposable  
(immutable/read-only) VMs?

It came up at Allmydata in the context of installer testing, but I've  
had two or three conversations in the last month or so where it would  
be desirable to have a build happen on a pristine VM.

My thoughts were along the lines of having a 'controller' buildslave  
that responds to changes by starting a new VM from an immutable image,  
and a buildslave inside that VM that starts at boot, connects, and  
finds work.

Cleanup at the other end is a little less clear, so I'm looking for  
suggestions there.

Thanks,
	Zandr

PS: Stephen- Thanks for the pointer to Loki, that might make my life a  
lot easier.



From gary.poster at canonical.com  Tue Feb 10 16:28:59 2009
From: gary.poster at canonical.com (Gary Poster)
Date: Tue, 10 Feb 2009 11:28:59 -0500
Subject: [Buildbot-devel] Disposable VM Slaves?
In-Reply-To: <1F7FCC76-0499-4931-A201-C3E8F944FCEB@allmydata.com>
References: <1F7FCC76-0499-4931-A201-C3E8F944FCEB@allmydata.com>
Message-ID: <1234283339.7262.7624.camel@gary-ubuntu>

On Tue, 2009-02-10 at 08:14 -0800, Aleksandr Milewski wrote:
> Does anyone have a good solution for running buildslaves on disposable  
> (immutable/read-only) VMs?
> 
> It came up at Allmydata in the context of installer testing, but I've  
> had two or three conversations in the last month or so where it would  
> be desirable to have a build happen on a pristine VM.
> 
> My thoughts were along the lines of having a 'controller' buildslave  
> that responds to changes by starting a new VM from an immutable image,  
> and a buildslave inside that VM that starts at boot, connects, and  
> finds work.
> 
> Cleanup at the other end is a little less clear, so I'm looking for  
> suggestions there.

It sounds like you might be interested in the EC2LatentBuildSlave I
added recently.  It's in Dustin's master branch on github.  If you don't
want to use EC2, you can subclass the abstract implementation
(AbstractLatentBuildSlave) and follow the basic pattern.

Gary




From seefeld at sympatico.ca  Tue Feb 10 16:29:58 2009
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 10 Feb 2009 11:29:58 -0500
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>
Message-ID: <4991AB86.80809@sympatico.ca>

Aleksandr Milewski wrote:
> Agreed, this is very high on my list.
>
> I'd also like to be able to subscribe a single buildslave to multiple  
> buildmasters.
>   

Out of curiosity, what is the rationale for this ? While I can see it 
being hard to maintain multiple buildmasters, build slaves are cheap, 
and allowing them to serve multiple projects just complicates things 
without much benefit.

FWIW.

Thanks,
       Stefan


-- 

      ...ich hab' noch einen Koffer in Berlin...




From l10n.moz at googlemail.com  Tue Feb 10 16:44:00 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Tue, 10 Feb 2009 17:44:00 +0100
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <4991AB86.80809@sympatico.ca>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>
	<4991AB86.80809@sympatico.ca>
Message-ID: <9de1bdb60902100844i9dff012g9d6d045900de21fa@mail.gmail.com>

2009/2/10 Stefan Seefeld <seefeld at sympatico.ca>

> Aleksandr Milewski wrote:
> > Agreed, this is very high on my list.
> >
> > I'd also like to be able to subscribe a single buildslave to multiple
> > buildmasters.
> >
>
> Out of curiosity, what is the rationale for this ? While I can see it
> being hard to maintain multiple buildmasters, build slaves are cheap,
> and allowing them to serve multiple projects just complicates things
> without much benefit.
>
>
Interestingly enough, build slaves aren't as cheap as one would think. Right
now, we're running against the limits of our VMs at a few places here at
Mozilla, and we're already at a size where just adding another machine
doesn't really solve the problem.

Another thing that I have in my mind is that I'm not sure if we're not
reaching the limits of what our masters can handle. And if the master would
like to be split, we'd still like to use one pool of slaves, and have locks
work.

Which is probably the hardest thing to do in the whole piece.

Axel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090210/a0ef3ad6/attachment.html>

From seefeld at sympatico.ca  Tue Feb 10 16:45:24 2009
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 10 Feb 2009 11:45:24 -0500
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <9FB32AAF-F248-4C2F-AE81-72FE4C3E453C@smadness.com>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>
	<4991AB86.80809@sympatico.ca>
	<9FB32AAF-F248-4C2F-AE81-72FE4C3E453C@smadness.com>
Message-ID: <4991AF24.9080106@sympatico.ca>

john gale wrote:
>
> On Feb 10, 2009, at 8:29 AM, Stefan Seefeld wrote:
>
>>> Agreed, this is very high on my list.
>>>
>>> I'd also like to be able to subscribe a single buildslave to multiple
>>> buildmasters.
>>>
>>
>> Out of curiosity, what is the rationale for this ? While I can see it
>> being hard to maintain multiple buildmasters, build slaves are cheap,
>> and allowing them to serve multiple projects just complicates things
>> without much benefit.
>
>
>
> Build slaves aren't always cheap.  Depending on what you're testing, 
> ideally you'd want one machine for each build slave.  Registering 
> multiple build slaves per machine can start running on each others 
> toes, especially when you're testing system software.

I see, and agree. However, that can already be dealt with on the build 
master side, i.e. via coordinated scheduling and locking.
This raises another interesting issue that I have brought up here on 
numerous times (but unfortunately without sufficient insight and time to 
address it myself): load balancing, i.e. the ability for the build 
master to pick among a set of 'equal' slaves, depending on their load. 
(Note that not all of the load may stem from builders, so it is 
important to monitor resource usage that's coming from external tasks 
not under control of a buildbot instance.

Thanks,
       Stefan


-- 

      ...ich hab' noch einen Koffer in Berlin...




From john at smadness.com  Tue Feb 10 16:32:01 2009
From: john at smadness.com (john gale)
Date: Tue, 10 Feb 2009 08:32:01 -0800
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <4991AB86.80809@sympatico.ca>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>
	<4991AB86.80809@sympatico.ca>
Message-ID: <9FB32AAF-F248-4C2F-AE81-72FE4C3E453C@smadness.com>


On Feb 10, 2009, at 8:29 AM, Stefan Seefeld wrote:

>> Agreed, this is very high on my list.
>>
>> I'd also like to be able to subscribe a single buildslave to multiple
>> buildmasters.
>>
>
> Out of curiosity, what is the rationale for this ? While I can see it
> being hard to maintain multiple buildmasters, build slaves are cheap,
> and allowing them to serve multiple projects just complicates things
> without much benefit.



Build slaves aren't always cheap.  Depending on what you're testing,  
ideally you'd want one machine for each build slave.  Registering  
multiple build slaves per machine can start running on each others  
toes, especially when you're testing system software.

	~ john

=================
support the arts!
www.smadness.com

updated October 12, 2005





From john at smadness.com  Tue Feb 10 16:52:28 2009
From: john at smadness.com (john gale)
Date: Tue, 10 Feb 2009 08:52:28 -0800
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <4991AF24.9080106@sympatico.ca>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>
	<4991AB86.80809@sympatico.ca>
	<9FB32AAF-F248-4C2F-AE81-72FE4C3E453C@smadness.com>
	<4991AF24.9080106@sympatico.ca>
Message-ID: <023F2F8C-0FFF-4331-9910-E25D7F4641F9@smadness.com>


On Feb 10, 2009, at 8:45 AM, Stefan Seefeld wrote:

>>>> I'd also like to be able to subscribe a single buildslave to  
>>>> multiple
>>>> buildmasters.
>>>
>>> Out of curiosity, what is the rationale for this ? While I can see  
>>> it
>>> being hard to maintain multiple buildmasters, build slaves are  
>>> cheap,
>>> and allowing them to serve multiple projects just complicates things
>>> without much benefit.
>>
>> Build slaves aren't always cheap.  Depending on what you're  
>> testing, ideally you'd want one machine for each build slave.   
>> Registering multiple build slaves per machine can start running on  
>> each others toes, especially when you're testing system software.
>
> I see, and agree. However, that can already be dealt with on the  
> build master side, i.e. via coordinated scheduling and locking.



But that's my point, making sure the build master knows the status of  
the build slave only occurs once per master control program.  To get  
the benefits of many, many projects running (scheduled, without  
stepping on each others' toes) on a single build slave, they all have  
to be scheduled and locked by one master.  Which then means that you  
run into the original statement of cluttering one master with many  
disparate projects.

I would indeed love to see the ability to have a pool of slaves, that  
can license themselves out to different masters, fully coordinated and  
lockable, so that many teams could implement their own buildbot master  
and share the same slave resources.

	~ john

=================
support the arts!
www.smadness.com

updated October 12, 2005





From seefeld at sympatico.ca  Tue Feb 10 16:54:47 2009
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 10 Feb 2009 11:54:47 -0500
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <9de1bdb60902100844i9dff012g9d6d045900de21fa@mail.gmail.com>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>	
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>	
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>	
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>	
	<4991AB86.80809@sympatico.ca>
	<9de1bdb60902100844i9dff012g9d6d045900de21fa@mail.gmail.com>
Message-ID: <4991B157.9030307@sympatico.ca>

Axel Hecht wrote:
> 2009/2/10 Stefan Seefeld <seefeld at sympatico.ca 
> <mailto:seefeld at sympatico.ca>>
>
>     Aleksandr Milewski wrote:
>     > Agreed, this is very high on my list.
>     >
>     > I'd also like to be able to subscribe a single buildslave to
>     multiple
>     > buildmasters.
>     >
>
>     Out of curiosity, what is the rationale for this ? While I can see it
>     being hard to maintain multiple buildmasters, build slaves are cheap,
>     and allowing them to serve multiple projects just complicates things
>     without much benefit.
>
>  
> Interestingly enough, build slaves aren't as cheap as one would think. 
> Right now, we're running against the limits of our VMs at a few places 
> here at Mozilla, and we're already at a size where just adding another 
> machine doesn't really solve the problem.

OK, it seems I expressed myself poorly. I was certainly not suggesting 
that a build slave instance, executing one or multiple builders, was 
cheap. My point was that I couldn't see the advantage of one slave 
serving multiple builders / masters.
I fully agree that it is highly desirable to be able to restrict how 
many builders may run in parallel.

Thanks,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...




From john at smadness.com  Tue Feb 10 16:58:02 2009
From: john at smadness.com (john gale)
Date: Tue, 10 Feb 2009 08:58:02 -0800
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <4991B157.9030307@sympatico.ca>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>
	<4991AB86.80809@sympatico.ca>
	<9de1bdb60902100844i9dff012g9d6d045900de21fa@mail.gmail.com>
	<4991B157.9030307@sympatico.ca>
Message-ID: <94DDB94C-4E54-4019-A8B9-07749AA65BE9@smadness.com>


On Feb 10, 2009, at 8:54 AM, Stefan Seefeld wrote:

> OK, it seems I expressed myself poorly. I was certainly not suggesting
> that a build slave instance, executing one or multiple builders, was
> cheap. My point was that I couldn't see the advantage of one slave
> serving multiple builders / masters.



At Apple we could have a pool of slaves...  one configuration of  
anything ever (recently) shipped.  This would be an insane collection  
of machines, unreasonable to replicate through different teams...  but  
if they could all be part of a slave pool assignable to multiple  
teams' masters, we could have, say, thirty different projects that  
each get a small slice of time on every config.  That would be a huge  
boon to our coverage...

	~ john

=================
support the arts!
www.smadness.com

updated October 12, 2005





From dustin at zmanda.com  Tue Feb 10 17:47:57 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Tue, 10 Feb 2009 12:47:57 -0500
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <49917052.80706@starswan.com>
References: <49917052.80706@starswan.com>
Message-ID: <42338fbf0902100947j4da6306fxb3ffb8bea7e00db6@mail.gmail.com>

First, to the others who have contributed feature requests on this
thread -- thank you!  Most of the features discussed are planned for
the 1.0 release.  We need more developers looking at this branch and
hacking away on it.  Neil has already started, and is doing great
things.

On Tue, Feb 10, 2009 at 7:17 AM, Stephen Dicks <stephen at starswan.com> wrote:
>    Having introduced buildbot to several companies over the last few
> years, right now I find myself in what some contractors call 'on the
> bench' (ie looking for work). Buildbot seems to have made quite slow
> progress recently (I started with 0.7.4 in 2005), so I would like to
> give it a 'shot in the arm' if I can - but I would like some advice as
> to which areas I should tackle. I can see that there are a large number
> of trac tickets outstanding and also the 'shimmering vision' of BuildBot
> 1.0 - but I'm not sure where to start.

This is great!  You should start wherever you feel most comfortable,
but ideally you can take on some management of the project, in
addition to contributing code.  The current management tasks are:
 - get 0.7.10 out the door
   - triage bugs in the "undecided" milestone
   - solve outstanding blocker bugs
   - make a release candidate
 - continue to manage 0.7.x releases
 - hack on and eventually release 1.0
 - establish a core group of committers to review and commit submissions
 - re-connect trac and metabuildbot to the source on github
 - triage the bugs on the SF tracker
   - move relevant bugs to trac
   - shut down SF tracker
 - remove SF CVS repository
 - remove Darcs repository

Currently, I'm focused on the first point (0.7.10 release), and
itching to get back to hacking on the 1.0 branch.  I don't anticipate
having enough time to accomplish the remaining tasks, so any offers of
assistance are greatly appreciated.  We need a *team* to run this
project.

I'll be happy to talk off-list about specific tasks, if you'd like.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From zandr at allmydata.com  Tue Feb 10 19:17:07 2009
From: zandr at allmydata.com (Aleksandr Milewski)
Date: Tue, 10 Feb 2009 11:17:07 -0800
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <4991B157.9030307@sympatico.ca>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>	
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>	
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>	
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>	
	<4991AB86.80809@sympatico.ca>
	<9de1bdb60902100844i9dff012g9d6d045900de21fa@mail.gmail.com>
	<4991B157.9030307@sympatico.ca>
Message-ID: <368A1500-BFDF-4129-B356-95D10F60662B@allmydata.com>


On Feb 10, 2009, at 8:54 AM, Stefan Seefeld wrote:
> OK, it seems I expressed myself poorly. I was certainly not  
> suggesting that a build slave instance, executing one or multiple  
> builders, was cheap. My point was that I couldn't see the advantage  
> of one slave serving multiple builders / masters.
> I fully agree that it is highly desirable to be able to restrict how  
> many builders may run in parallel.

When considering this, keep in mind that it may be desirable for a  
slave to run builds for different masters run by different teams. I  
have a slave running a few builders for Allmydata on my amd64 box at  
home. I'd like to contribute cycles on that box to other open source  
projects as well, but I really don't want more than one or two  
builders running at once.

In chatting with Brian about this quite a while ago, he seemed to  
think that running a single slave process that polled multiple masters  
looking for work would be the easiest way to to implement this.

However, doing coordination in some way between multiple slaves on a  
single box would meet my requirements. Just don't assume that  
coordination between masters to achieve this is possible.

-Zandr



From seefeld at sympatico.ca  Tue Feb 10 19:25:23 2009
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 10 Feb 2009 14:25:23 -0500
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <368A1500-BFDF-4129-B356-95D10F60662B@allmydata.com>
References: <1657625930.5540781234280861816.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>	
	<1317657598.5540981234280903900.JavaMail.root@zmail01.collab.prod.int.phx2.redhat.com>	
	<77fcfd0a0902100756p2e04375eq88ef6076f46df047@mail.gmail.com>	
	<03036A22-89DF-4DA9-830A-6CD913E13A44@allmydata.com>	
	<4991AB86.80809@sympatico.ca>
	<9de1bdb60902100844i9dff012g9d6d045900de21fa@mail.gmail.com>
	<4991B157.9030307@sympatico.ca>
	<368A1500-BFDF-4129-B356-95D10F60662B@allmydata.com>
Message-ID: <4991D4A3.5070007@sympatico.ca>

Aleksandr Milewski wrote:
>
> On Feb 10, 2009, at 8:54 AM, Stefan Seefeld wrote:
>> OK, it seems I expressed myself poorly. I was certainly not 
>> suggesting that a build slave instance, executing one or multiple 
>> builders, was cheap. My point was that I couldn't see the advantage 
>> of one slave serving multiple builders / masters.
>> I fully agree that it is highly desirable to be able to restrict how 
>> many builders may run in parallel.
>
> When considering this, keep in mind that it may be desirable for a 
> slave to run builds for different masters run by different teams. I 
> have a slave running a few builders for Allmydata on my amd64 box at 
> home. I'd like to contribute cycles on that box to other open source 
> projects as well, but I really don't want more than one or two 
> builders running at once.
>
> In chatting with Brian about this quite a while ago, he seemed to 
> think that running a single slave process that polled multiple masters 
> looking for work would be the easiest way to to implement this.
>
> However, doing coordination in some way between multiple slaves on a 
> single box would meet my requirements. Just don't assume that 
> coordination between masters to achieve this is possible.
I believe you are making my point. A long time ago I proposed load 
balancing to address this (http://buildbot.net/trac/ticket/42), 
precisely because I found it too limiting to assume that all the 
processes running on the machine were controlled by some buildbot master 
that one could coordinate with. (For example, using os.getloadavg() 
where available.)

Thanks,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...




From smilner at redhat.com  Tue Feb 10 20:09:52 2009
From: smilner at redhat.com (Steve 'Ashcrow' Milner)
Date: Tue, 10 Feb 2009 15:09:52 -0500
Subject: [Buildbot-devel] Full-Time buildbot developer available...!
In-Reply-To: <42338fbf0902100947j4da6306fxb3ffb8bea7e00db6@mail.gmail.com>
References: <49917052.80706@starswan.com>
	<42338fbf0902100947j4da6306fxb3ffb8bea7e00db6@mail.gmail.com>
Message-ID: <20090210200952.GB24462@arc.rdu.redhat.com>

On 10/02/09 12:47 -0500, Dustin J. Mitchell wrote:
>Currently, I'm focused on the first point (0.7.10 release), and
>itching to get back to hacking on the 1.0 branch.  I don't anticipate
>having enough time to accomplish the remaining tasks, so any offers of
>assistance are greatly appreciated.  We need a *team* to run this
>project.

I'll try to jump in on a ticket or two for 0.7.10.

-- 
kthxbye!
Steve 'Ashcrow' Milner
Team: IT Engineering Support Lead
Location: 245 Centennial
IRC: ashcrow
GnuPG ID: 28DFD4BE

"In the heat of conversation I may have said certain things I believe 
to be untrue. The alleged lie that you might have heard me saying 
allegedly moments ago ... thats a parasite that lives in my neck." 
     -- Tad Ghostal
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://buildbot.net/pipermail/devel/attachments/20090210/4e962aac/attachment.bin>

From mahatmamanic at gmail.com  Tue Feb 10 23:39:51 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Tue, 10 Feb 2009 15:39:51 -0800
Subject: [Buildbot-devel] Properties objects
Message-ID: <c02dbb2a0902101539p26dbfde7w7d4ce34dba017114@mail.gmail.com>

If in an extended class I want to pass a custom Properties object to a
BuildRequest constructor, is Properties.update the correct way to do that,
like so:

d = {}
d['f'oo'] = bar
d['baz'] = quz
p = Properties()
p.update(d, "MY STUFF")

request = BuildRequest(reason, source, builder, p)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090210/a1d161d3/attachment.html>

From dustin at zmanda.com  Wed Feb 11 00:02:35 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Tue, 10 Feb 2009 19:02:35 -0500
Subject: [Buildbot-devel] DirectoryUpload
In-Reply-To: <gmq4s5$oul$1@ger.gmane.org>
References: <ggm65p$uo6$1@ger.gmane.org> <492ED4AF.7080202@yar.nu>
	<ggo8au$bht$1@ger.gmane.org> <492FA7FA.5060307@yar.nu>
	<gh3g6v$kv5$1@ger.gmane.org> <493597A2.30503@yar.nu>
	<gmq4s5$oul$1@ger.gmane.org>
Message-ID: <42338fbf0902101602x6bb4548cvd86fe24650d6c95e@mail.gmail.com>

On Mon, Feb 9, 2009 at 3:49 PM, Ren? M?ller <mue.r at gmx.de> wrote:
> I completely remade the patch and now it works great. I added some
> tests, too. What documentation is needed to get a review?

I wrote on the ticket that I can't get this into 0.7.10, but that's
not really true.  If you can add a patch for docs/buildbot.texinfo to
explain how to use your new step, and some unit tests to make sure it
works (see test_transfer.py), we can get it in.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Wed Feb 11 04:45:39 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Tue, 10 Feb 2009 23:45:39 -0500
Subject: [Buildbot-devel] Overriding default IRC behavior
In-Reply-To: <c02dbb2a0902091736m212af484k1fbf6b187679ea2f@mail.gmail.com>
References: <c02dbb2a0902091736m212af484k1fbf6b187679ea2f@mail.gmail.com>
Message-ID: <42338fbf0902102045k4fade766q86d85c05f545e2cc@mail.gmail.com>

On Mon, Feb 9, 2009 at 8:36 PM, Ian Peters-Campbell
<mahatmamanic at gmail.com> wrote:
> Has anyone done anything similar and does that seem like a reasonable
> approach, or is there a pre-existing trick or a way to write less code that
> will let my teammates specify things like target servers or signing
> credentials when requesting a build?

I can't comment on the particular proposal, but you should note that
there's been a lot of work on the IRC interface since the last
release, so you should be sure to base your work on the latest work in
the git repository.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From al4321 at gmail.com  Wed Feb 11 12:44:21 2009
From: al4321 at gmail.com (Alexey Eremenko)
Date: Wed, 11 Feb 2009 12:44:21 +0000
Subject: [Buildbot-devel] Why 3-level separation?
Message-ID: <7fac565a0902110444t383647c6y4cfda58eb774cddc@mail.gmail.com>

Hi All !

I have read the docs, but I am still struggling to understand why
BuildBot has 3 level hierarchy (factory/builder/build-slave) ?

Why two levels hierarchy is not enough (factory/build-slave) ?

-- 
-Alexey Eromenko "Technologov"



From l10n.moz at googlemail.com  Wed Feb 11 13:33:04 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Wed, 11 Feb 2009 14:33:04 +0100
Subject: [Buildbot-devel] Why 3-level separation?
In-Reply-To: <7fac565a0902110444t383647c6y4cfda58eb774cddc@mail.gmail.com>
References: <7fac565a0902110444t383647c6y4cfda58eb774cddc@mail.gmail.com>
Message-ID: <9de1bdb60902110533t54a75ffax42ebd377ad7a593b@mail.gmail.com>

2009/2/11 Alexey Eremenko <al4321 at gmail.com>

> Hi All !
>
> I have read the docs, but I am still struggling to understand why
> BuildBot has 3 level hierarchy (factory/builder/build-slave) ?
>
> Why two levels hierarchy is not enough (factory/build-slave) ?
>

You can use the same factory for builders on different platforms.

A factory describes what the build does, a builder describes which of those
builds you do, and each has a bunch of buildslaves to actually execute those
builds.

Axel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090211/862c2603/attachment.html>

From dustin at zmanda.com  Wed Feb 11 13:41:43 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 11 Feb 2009 08:41:43 -0500
Subject: [Buildbot-devel] Why 3-level separation?
In-Reply-To: <7fac565a0902110444t383647c6y4cfda58eb774cddc@mail.gmail.com>
References: <7fac565a0902110444t383647c6y4cfda58eb774cddc@mail.gmail.com>
Message-ID: <42338fbf0902110541y46febd38r5910f3c65204717c@mail.gmail.com>

On Wed, Feb 11, 2009 at 7:44 AM, Alexey Eremenko <al4321 at gmail.com> wrote:
> I have read the docs, but I am still struggling to understand why
> BuildBot has 3 level hierarchy (factory/builder/build-slave) ?
>
> Why two levels hierarchy is not enough (factory/build-slave) ?

I'm not sure there's a good answer to that, except that it was
designed that way.  Having an extra level of flexibility doesn't hurt.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Wed Feb 11 13:48:05 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 11 Feb 2009 08:48:05 -0500
Subject: [Buildbot-devel] Properties objects
In-Reply-To: <c02dbb2a0902101539p26dbfde7w7d4ce34dba017114@mail.gmail.com>
References: <c02dbb2a0902101539p26dbfde7w7d4ce34dba017114@mail.gmail.com>
Message-ID: <42338fbf0902110548p6c350766nf0e5112f42b61a73@mail.gmail.com>

On Tue, Feb 10, 2009 at 6:39 PM, Ian Peters-Campbell
<mahatmamanic at gmail.com> wrote:
> d = {}
> d['f'oo'] = bar
> d['baz'] = quz
> p = Properties()
> p.update(d, "MY STUFF")
>
> request = BuildRequest(reason, source, builder, p)

Probably easier to write
request = BuildRequest(reason, source, builder,
  Properties(foo=bar, baz=quz))

-- 
Storage Software Engineer
http://www.zmanda.com



From mahatmamanic at gmail.com  Wed Feb 11 14:13:37 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Wed, 11 Feb 2009 06:13:37 -0800
Subject: [Buildbot-devel] Why 3-level separation?
In-Reply-To: <42338fbf0902110541y46febd38r5910f3c65204717c@mail.gmail.com>
References: <7fac565a0902110444t383647c6y4cfda58eb774cddc@mail.gmail.com>
	<42338fbf0902110541y46febd38r5910f3c65204717c@mail.gmail.com>
Message-ID: <c02dbb2a0902110613r2d0e8bev640b1afba9c8c191@mail.gmail.com>

I can't speak to why it was written the way it was, but I can tell you why I
like it. The factory, builder, and build-slave each answer different
questions about the build process.

Buildslaves tell you Where
Factories tell you How
Builds tell you What
Build Requests tell you Who (through manual) or When (through Schedulers),
and Why

Since the different questions about the build are broken into different
objects it allows you to be more efficient both in writing code and in using
physical resources.

It also keeps things more sane by ensuring that particular information about
your build system/builds only needs to be represented once.  By splitting up
the individual questions about the build there is no need to duplicate the
code for build steps, build slave location, SVN branch info, etc.  That lets
you make fewer code/configuration changes and avoid frustrating problems
down the road.

My .02 :)

Ian


On Wed, Feb 11, 2009 at 5:41 AM, Dustin J. Mitchell <dustin at zmanda.com>wrote:

> On Wed, Feb 11, 2009 at 7:44 AM, Alexey Eremenko <al4321 at gmail.com> wrote:
> > I have read the docs, but I am still struggling to understand why
> > BuildBot has 3 level hierarchy (factory/builder/build-slave) ?
> >
> > Why two levels hierarchy is not enough (factory/build-slave) ?
>
> I'm not sure there's a good answer to that, except that it was
> designed that way.  Having an extra level of flexibility doesn't hurt.
>
> Dustin
>
> --
> Storage Software Engineer
> http://www.zmanda.com
>
>
> ------------------------------------------------------------------------------
> Create and Deploy Rich Internet Apps outside the browser with
> Adobe(R)AIR(TM)
> software. With Adobe AIR, Ajax developers can use existing skills and code
> to
> build responsive, highly engaging applications that combine the power of
> local
> resources and data with the reach of the web. Download the Adobe AIR SDK
> and
> Ajax docs to start building applications today-
> http://p.sf.net/sfu/adobe-com
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090211/3fcc0792/attachment.html>

From gerg.ward+buildbot at gmail.com  Wed Feb 11 13:48:35 2009
From: gerg.ward+buildbot at gmail.com (Greg Ward)
Date: Wed, 11 Feb 2009 08:48:35 -0500
Subject: [Buildbot-devel] Why 3-level separation?
In-Reply-To: <7fac565a0902110444t383647c6y4cfda58eb774cddc@mail.gmail.com>
References: <7fac565a0902110444t383647c6y4cfda58eb774cddc@mail.gmail.com>
Message-ID: <20090211134835.GC29294@intelerad.com>

On 11 February 2009, Alexey Eremenko said:
> I have read the docs, but I am still struggling to understand why
> BuildBot has 3 level hierarchy (factory/builder/build-slave) ?

Actually I think a better way to look at the hierarchy is:

  (factory/builder) -> build -> build step

where the build step can (but does not have to) run code remotely,
on the slave.  (It always runs code on the master, starting in start()
and ending in finish().)

The important difference is lifetime.  The factory and builder are
both long-lived: I like to think of a complete column in the waterfall
view, from the beginning of time to right now, as a "builder".  (It's
a little more complicated than that because some of that info lives on
disk and only some of it is in memory, but never mind that.)

Each "build", on the other hand, is spawned from a builder, and has a
limited lifetime.  If it takes 10 min to checkout/configure/make/test
your project, then the corresponding build object lives for 10 min.

Finally, a build step is one component of the build.  It sticks around
for however long it takes to do the checkout, or make, or whatever.
Build steps are created from the build's builder's factory.  That's
where the factory comes into things.

       Greg



From gerg.ward+buildbot at gmail.com  Wed Feb 11 13:42:54 2009
From: gerg.ward+buildbot at gmail.com (Greg Ward)
Date: Wed, 11 Feb 2009 08:42:54 -0500
Subject: [Buildbot-devel] Properties objects
In-Reply-To: <c02dbb2a0902101539p26dbfde7w7d4ce34dba017114@mail.gmail.com>
References: <c02dbb2a0902101539p26dbfde7w7d4ce34dba017114@mail.gmail.com>
Message-ID: <20090211134254.GB29294@intelerad.com>

On 10 February 2009, Ian Peters-Campbell said:
> If in an extended class I want to pass a custom Properties object to a
> BuildRequest constructor, is Properties.update the correct way to do that,
> like so:
> 
> d = {}
> d['f'oo'] = bar
> d['baz'] = quz
> p = Properties()
> p.update(d, "MY STUFF")
> 
> request = BuildRequest(reason, source, builder, p)

Disclaimer: I have only read the code and scanned the docs, not
actually tried this (yet).  But it looks to me as if you could also do
that in your master.cfg:

  c['properties'] = { 'foo': bar, 'baz': qux }

and then the 'source' of those properties will be set automatically.

Depends on how static your property values are, of course.  I have
properties for an external build ID and build label, so those will
have to set when I launch each build.  Luckily I have custom code for
that (reads a record from our build queue and turns it into a Buildbot
BuildRequest), so it looks fairly straightforward.  (Fingers crossed.)

       Greg



From stephen at starswan.com  Wed Feb 11 17:47:36 2009
From: stephen at starswan.com (Stephen Dicks)
Date: Wed, 11 Feb 2009 17:47:36 +0000
Subject: [Buildbot-devel] The Meta Buildbot
Message-ID: <49930F38.2040400@starswan.com>

Does anyone know anything about the configuration of the meta-buildbot? 
Is it stored anywhere? If BB is going to make any progress then I think 
its quite an important thing to have. If I don't hear anything in the 
next few days I will probably set up one myself and point it at the main 
git repositories.

I seem to have found recently that the 'Trial' step of buildbot is 
broken under twisted 8.x (ubuntu intrepid ibis).

--
Steve Dicks



From dustin at zmanda.com  Wed Feb 11 18:01:10 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 11 Feb 2009 13:01:10 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <49930F38.2040400@starswan.com>
References: <49930F38.2040400@starswan.com>
Message-ID: <42338fbf0902111001w2195bfc7o6c8846d65f65cb5a@mail.gmail.com>

On Wed, Feb 11, 2009 at 12:47 PM, Stephen Dicks <stephen at starswan.com> wrote:
> Does anyone know anything about the configuration of the meta-buildbot?
> Is it stored anywhere? If BB is going to make any progress then I think
> its quite an important thing to have. If I don't hear anything in the
> next few days I will probably set up one myself and point it at the main
> git repositories.

There's been some discussion of this recently.  The current
metabuildbot is hosted on Brian's systems, to which I don't have
access just yet.  I would *love* to have you set this up, particularly
if it's configured in some publicly-accessible fashion, and allows
others to submit buildslaves.

I don't think SourceForge can host a long-running daemon like this, so
I imagine that someone will need to host the buildmaster.  I can do
this on my systems, but those are behind a cable modem and thus not
terribly reliable.  If someone has a bit of business bandwidth to
share for the purpose, that would be better.

We should probably set up a CNAME that the buildslaves connect to, so
that if we must move the master to another host, the buildslaves need
not be reconfigured (just restarted).  I think Brian can allocate a
subdomain of buildbot.net to this purpose.

The buildbot should watch the github repository, and preferably only
the 'master' branch (and maybe 'one-oh').  Let's also put the
configuration in a branch on github.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From stephen at starswan.com  Wed Feb 11 18:09:18 2009
From: stephen at starswan.com (Stephen Dicks)
Date: Wed, 11 Feb 2009 18:09:18 +0000
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <42338fbf0902111001w2195bfc7o6c8846d65f65cb5a@mail.gmail.com>
References: <49930F38.2040400@starswan.com>
	<42338fbf0902111001w2195bfc7o6c8846d65f65cb5a@mail.gmail.com>
Message-ID: <4993144E.80805@starswan.com>

Dustin J. Mitchell wrote:
> metabuildbot is hosted on Brian's systems, to which I don't have
> access just yet.  I would *love* to have you set this up, particularly
> if it's configured in some publicly-accessible fashion, and allows
> others to submit buildslaves.
>   
Sounds like another mini-project...:-)
> I don't think SourceForge can host a long-running daemon like this, so
> I imagine that someone will need to host the buildmaster.  I can do
> this on my systems, but those are behind a cable modem and thus not
> terribly reliable.  If someone has a bit of business bandwidth to
> share for the purpose, that would be better.
>   
I can provide that no problem.
> We should probably set up a CNAME that the buildslaves connect to, so
> that if we must move the master to another host, the buildslaves need
> not be reconfigured (just restarted).  I think Brian can allocate a
> subdomain of buildbot.net to this purpose.
>
> The buildbot should watch the github repository, and preferably only
> the 'master' branch (and maybe 'one-oh').  Let's also put the
> configuration in a branch on github.
>   
sounds like a plan. We could really do with some buildslave machines - I 
don't have access to MacOS but I have lots of PC hardware (noise 
permitting that is)



From deejay1 at srem.org  Wed Feb 11 18:11:19 2009
From: deejay1 at srem.org (=?UTF-8?B?xYF1a2FzeiBKZXJuYcWb?=)
Date: Wed, 11 Feb 2009 19:11:19 +0100
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <42338fbf0902111001w2195bfc7o6c8846d65f65cb5a@mail.gmail.com>
References: <49930F38.2040400@starswan.com>
	<42338fbf0902111001w2195bfc7o6c8846d65f65cb5a@mail.gmail.com>
Message-ID: <77fcfd0a0902111011i61f91402h6ef589d89f113b0e@mail.gmail.com>

On Wed, Feb 11, 2009 at 7:01 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
> On Wed, Feb 11, 2009 at 12:47 PM, Stephen Dicks <stephen at starswan.com> wrote:
>> Does anyone know anything about the configuration of the meta-buildbot?
>> Is it stored anywhere? If BB is going to make any progress then I think
>> its quite an important thing to have. If I don't hear anything in the
>> next few days I will probably set up one myself and point it at the main
>> git repositories.
>
> There's been some discussion of this recently.  The current
> metabuildbot is hosted on Brian's systems, to which I don't have
> access just yet.  I would *love* to have you set this up, particularly
> if it's configured in some publicly-accessible fashion, and allows
> others to submit buildslaves.
>
> I don't think SourceForge can host a long-running daemon like this, so
> I imagine that someone will need to host the buildmaster.  I can do
> this on my systems, but those are behind a cable modem and thus not
> terribly reliable.  If someone has a bit of business bandwidth to
> share for the purpose, that would be better.

Well, I have an Intel Atom lying around doing mostly nothing with
Ubuntu Hardy LTS on it and a 100mbps line (hopefully), I can spare one
or two user accounts on it. It is running BOINC mostly ;)
Feel free to contact me off list for that.

-- 
?ukasz



From mahatmamanic at gmail.com  Wed Feb 11 23:18:01 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Wed, 11 Feb 2009 15:18:01 -0800
Subject: [Buildbot-devel] custom build steps
Message-ID: <c02dbb2a0902111518u262fb6fbx909eaa53bbebdd7d@mail.gmail.com>

Is there a form of existing BuildStep which, assuming I have Python and the
relevant libraries installed on the BuildSlave, would let me run the Python
code from the build step's start method (or some other arbitrary function)
locally to the build slave rather than the build master?  So for instance,
if I had code such as:

f = open('foo.txt', 'w')
f.write("bar")
f.close()

I would want the file foo.txt to appear on the build slave rather than on
the build master.

If not I guess I can write my code to a file, send the file down with a
FileDownload, and then execute it with a ShellCommand, but that feels like
the long way around the lake, and I figured I'd ask before I went with that.

Thanks
Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090211/6c9a8626/attachment.html>

From markroddy at gmail.com  Wed Feb 11 23:42:57 2009
From: markroddy at gmail.com (Mark Roddy)
Date: Wed, 11 Feb 2009 18:42:57 -0500
Subject: [Buildbot-devel] custom build steps
In-Reply-To: <c02dbb2a0902111518u262fb6fbx909eaa53bbebdd7d@mail.gmail.com>
References: <c02dbb2a0902111518u262fb6fbx909eaa53bbebdd7d@mail.gmail.com>
Message-ID: <7930f1200902111542u67c9d63ele9f61909570efaee@mail.gmail.com>

Your example is obviously a simplified version of whatever it is you'd
like to accomplish so I can't be sure that I am under the correct
impression of your goal, but if you're trying to run your build via
buildbot I think you should consider keeping the python code you want
to execute under source control.  I try to keep any code needed for
the build process in source control and not in buildbot itself to
ensure that any dev on the team can perform all necessary steps for
the build on their local machine.  Unless there's something unique to
what you're trying to accomplish, this is what I would advise.

-Mark


On Wed, Feb 11, 2009 at 6:18 PM, Ian Peters-Campbell
<mahatmamanic at gmail.com> wrote:
> Is there a form of existing BuildStep which, assuming I have Python and the
> relevant libraries installed on the BuildSlave, would let me run the Python
> code from the build step's start method (or some other arbitrary function)
> locally to the build slave rather than the build master?  So for instance,
> if I had code such as:
>
> f = open('foo.txt', 'w')
> f.write("bar")
> f.close()
>
> I would want the file foo.txt to appear on the build slave rather than on
> the build master.
>
> If not I guess I can write my code to a file, send the file down with a
> FileDownload, and then execute it with a ShellCommand, but that feels like
> the long way around the lake, and I figured I'd ask before I went with that.
>
> Thanks
> Ian
>
>
> ------------------------------------------------------------------------------
> Create and Deploy Rich Internet Apps outside the browser with
> Adobe(R)AIR(TM)
> software. With Adobe AIR, Ajax developers can use existing skills and code
> to
> build responsive, highly engaging applications that combine the power of
> local
> resources and data with the reach of the web. Download the Adobe AIR SDK and
> Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
>



From mue.r at gmx.de  Thu Feb 12 07:53:58 2009
From: mue.r at gmx.de (=?UTF-8?B?UmVuw6kgTcO8bGxlcg==?=)
Date: Thu, 12 Feb 2009 08:53:58 +0100
Subject: [Buildbot-devel] DirectoryUpload
In-Reply-To: <42338fbf0902101602x6bb4548cvd86fe24650d6c95e@mail.gmail.com>
References: <ggm65p$uo6$1@ger.gmane.org>
	<492ED4AF.7080202@yar.nu>	<ggo8au$bht$1@ger.gmane.org>
	<492FA7FA.5060307@yar.nu>	<gh3g6v$kv5$1@ger.gmane.org>
	<493597A2.30503@yar.nu>	<gmq4s5$oul$1@ger.gmane.org>
	<42338fbf0902101602x6bb4548cvd86fe24650d6c95e@mail.gmail.com>
Message-ID: <gn0kim$grs$1@ger.gmane.org>

Dustin J. Mitchell schrieb:
> On Mon, Feb 9, 2009 at 3:49 PM, Ren? M?ller <mue.r at gmx.de> wrote:
>> I completely remade the patch and now it works great. I added some
>> tests, too. What documentation is needed to get a review?
> 
> I wrote on the ticket that I can't get this into 0.7.10, but that's
> not really true.  If you can add a patch for docs/buildbot.texinfo to
> explain how to use your new step, and some unit tests to make sure it
> works (see test_transfer.py), we can get it in.
> 

My patch already adds some unit tests to test_transfer.py. I will try to
make a patch for docs/buildbot.texinfo over the weekend.

Ren?




From dustin at zmanda.com  Thu Feb 12 14:17:40 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Thu, 12 Feb 2009 09:17:40 -0500
Subject: [Buildbot-devel] DirectoryUpload
In-Reply-To: <gn0kim$grs$1@ger.gmane.org>
References: <ggm65p$uo6$1@ger.gmane.org> <492ED4AF.7080202@yar.nu>
	<ggo8au$bht$1@ger.gmane.org> <492FA7FA.5060307@yar.nu>
	<gh3g6v$kv5$1@ger.gmane.org> <493597A2.30503@yar.nu>
	<gmq4s5$oul$1@ger.gmane.org>
	<42338fbf0902101602x6bb4548cvd86fe24650d6c95e@mail.gmail.com>
	<gn0kim$grs$1@ger.gmane.org>
Message-ID: <42338fbf0902120617p21942ef3j21efba016e362354@mail.gmail.com>

On Thu, Feb 12, 2009 at 2:53 AM, Ren? M?ller <mue.r at gmx.de> wrote:
> My patch already adds some unit tests to test_transfer.py. I will try to
> make a patch for docs/buildbot.texinfo over the weekend.

Sorry, I don't know how I missed that :)

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From markus.kramer at mainconcept.de  Fri Feb 13 11:13:15 2009
From: markus.kramer at mainconcept.de (Markus Kramer)
Date: Fri, 13 Feb 2009 12:13:15 +0100
Subject: [Buildbot-devel] SlaveLocks
Message-ID: <499555CB.6050409@mainconcept.de>

Hi all,

i am just trying to setup some new buildmachines within our buildbot
system. The original idea was to have load sharing with some machines,
but somehow it does not work as expected.
I have setup multiple builders containing the following lines in their
build description:

'slavenames':[ "win-ia32-xp-b1","win-ia32-xp-b2"],
'locks': [xp_sl],


In addition i defined the lock like this in the top of config:

xp_sl = locks.SlaveLock("cpuxp2",maxCount=1)


I would have assumed that due to the fact that SlaveLocks are kept per
Slave, this would effectivly distribute the builds among these machines,
with only one build going on per Machine, but instead i still see
all the builds competing for access to machine b1.

Is the availability of locks not considered when selecting buildslave?
Is there probably some other method to make each build go to next free
slave and only queue when all of them are blocked?

Best regards,

Markus




-- 
   Markus Kramer
   Senior Engineer Codecs
   Multi-Platform Development Manager

   MainConcept GmbH
   Elisabethstr. 1
   52062 Aachen, Germany

   Phone: +49 (241) 40108-0
   Fax:     +49 (241) 40108-10
   Email:   markus.kramer at mainconcept.com
   Web:    www.mainconcept.com

   District Court Aachen: HRB 14788, Managing Directors: Muzaffer
Beygirci, David J. Richter, Eric Rodli


This message contains confidential, proprietary and/or attorney-client
privileged information or material owned by MainConcept and is intended
only for the addressed named individual(s).  The information contained
in this message may not be copied or distributed without express written
consent by an officer of MainConcept.  If the information contained in
this email is communicated under a confidentiality or non-disclosure
agreement with MainConcept, this email should be considered confidential
or proprietary written communication for the purposes of those
agreements.  Please notify the sender immediately by e-mail if you have
received this e-mail by mistake and delete this e-mail and related
information from your system.




From jarkko.palviainen at sesca.com  Fri Feb 13 12:19:55 2009
From: jarkko.palviainen at sesca.com (Jarkko Palviainen)
Date: Fri, 13 Feb 2009 14:19:55 +0200
Subject: [Buildbot-devel] Small correction to Buildbot manual (0.7.9)
Message-ID: <op.upahvhj5n4fw00@neppari.sesca.com>

Hi,

I was reading the maintenance chapter here:  
http://buildbot.net/repos/release/docs/buildbot.html#Maintenance and  
noticed a small glitch in the second crontab entry:

@weekly cd BASEDIR && find . -mindepth 2 -type f -mtime +14 -exec rm {} \;

The problem is that it actually removes files inside BASEDIR/public_html/  
directory if they have been laying around unmodified long enough (here,  
14*24 hours). At least it happened to me :)


A quick and dirty fix would be something like:
find . -mindepth 2 -type f -mtime +14 | grep -v public_html | xargs rm


Br,
Jarkko



From neil.hemingway at greyhavens.org.uk  Fri Feb 13 13:33:48 2009
From: neil.hemingway at greyhavens.org.uk (Neil Hemingway)
Date: Fri, 13 Feb 2009 13:33:48 +0000
Subject: [Buildbot-devel] SlaveLocks
In-Reply-To: <499555CB.6050409@mainconcept.de>
References: <499555CB.6050409@mainconcept.de>
Message-ID: <47be741a0902130533n7334878fmea1655b9afca8545@mail.gmail.com>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

That's exactly the config I use, and my builds get distributed across
my 3 slaves

Neil


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkmVdrwACgkQZjaLPuxR95R4lwCfd8llfWXzmP3DRU/T2cwJ944f
/9YAnRJUVwRSX1DqkDXImCjfyqCt8MV6
=lYbD
-----END PGP SIGNATURE-----
On 13/02/2009, Markus Kramer <markus.kramer at mainconcept.de> wrote:
>
> Hi all,
>
> i am just trying to setup some new buildmachines within our buildbot
> system. The original idea was to have load sharing with some machines,
> but somehow it does not work as expected.
> I have setup multiple builders containing the following lines in their
> build description:
>
> 'slavenames':[ "win-ia32-xp-b1","win-ia32-xp-b2"],
> 'locks': [xp_sl],
>
>
> In addition i defined the lock like this in the top of config:
>
> xp_sl = locks.SlaveLock("cpuxp2",maxCount=1)
>
>
> I would have assumed that due to the fact that SlaveLocks are kept per
> Slave, this would effectivly distribute the builds among these machines,
> with only one build going on per Machine, but instead i still see
> all the builds competing for access to machine b1.
>
> Is the availability of locks not considered when selecting buildslave?
> Is there probably some other method to make each build go to next free
> slave and only queue when all of them are blocked?
>
> Best regards,
>
> Markus
>
>
>
>
> --
>    Markus Kramer
>    Senior Engineer Codecs
>    Multi-Platform Development Manager
>
>    MainConcept GmbH
>    Elisabethstr. 1
>    52062 Aachen, Germany
>
>    Phone: +49 (241) 40108-0
>    Fax:     +49 (241) 40108-10
>    Email:   markus.kramer at mainconcept.com
>    Web:    www.mainconcept.com
>
>    District Court Aachen: HRB 14788, Managing Directors: Muzaffer
> Beygirci, David J. Richter, Eric Rodli
>
>
> This message contains confidential, proprietary and/or attorney-client
> privileged information or material owned by MainConcept and is intended
> only for the addressed named individual(s).  The information contained
> in this message may not be copied or distributed without express written
> consent by an officer of MainConcept.  If the information contained in
> this email is communicated under a confidentiality or non-disclosure
> agreement with MainConcept, this email should be considered confidential
> or proprietary written communication for the purposes of those
> agreements.  Please notify the sender immediately by e-mail if you have
> received this e-mail by mistake and delete this e-mail and related
> information from your system.
>
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
> CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the
> Enterprise
> -Strategies to boost innovation and cut costs with open source
> participation
> -Receive a $600 discount off the registration fee with the source code:
> SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090213/dfa15995/attachment.html>

From neil.hemingway at greyhavens.org.uk  Fri Feb 13 13:40:29 2009
From: neil.hemingway at greyhavens.org.uk (Neil Hemingway)
Date: Fri, 13 Feb 2009 13:40:29 +0000
Subject: [Buildbot-devel] Small correction to Buildbot manual (0.7.9)
In-Reply-To: <op.upahvhj5n4fw00@neppari.sesca.com>
References: <op.upahvhj5n4fw00@neppari.sesca.com>
Message-ID: <47be741a0902130540k1689e7farc3ef3b2a3c30d7cb@mail.gmail.com>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jarkko

This is covered by http://buildbot.net/trac/ticket/413 and
http://buildbot.net/trac/ticket/304.  A fix has been committed to
the main line of development and will be included in the next release
(0.7.10)

Neil


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkmVeE0ACgkQZjaLPuxR95Qx9QCgqy8zbhlwGkPGqHN3Osp9ayIM
SsoAn0DY18BmZczYzik0ACBkIVaZWPrS
=3pfd
-----END PGP SIGNATURE-----
On 13/02/2009, Jarkko Palviainen <jarkko.palviainen at sesca.com> wrote:
>
> Hi,
>
> I was reading the maintenance chapter here:
> http://buildbot.net/repos/release/docs/buildbot.html#Maintenance and
> noticed a small glitch in the second crontab entry:
>
> @weekly cd BASEDIR && find . -mindepth 2 -type f -mtime +14 -exec rm {} \;
>
> The problem is that it actually removes files inside BASEDIR/public_html/
> directory if they have been laying around unmodified long enough (here,
> 14*24 hours). At least it happened to me :)
>
>
> A quick and dirty fix would be something like:
> find . -mindepth 2 -type f -mtime +14 | grep -v public_html | xargs rm
>
>
> Br,
> Jarkko
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
> CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the
> Enterprise
> -Strategies to boost innovation and cut costs with open source
> participation
> -Receive a $600 discount off the registration fee with the source code:
> SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090213/6cea9450/attachment.html>

From markus.kramer at mainconcept.de  Fri Feb 13 14:15:30 2009
From: markus.kramer at mainconcept.de (Markus Kramer)
Date: Fri, 13 Feb 2009 15:15:30 +0100
Subject: [Buildbot-devel] SlaveLocks
In-Reply-To: <47be741a0902130533n7334878fmea1655b9afca8545@mail.gmail.com>
References: <499555CB.6050409@mainconcept.de>
	<47be741a0902130533n7334878fmea1655b9afca8545@mail.gmail.com>
Message-ID: <49958082.609@mainconcept.de>

Hi Neil,

so is there maybe some difference in versions of buildbot we are using. 
Mine should all be 0.7.9, but still do not properly share. I can see 
that i get a connect message for bot slaves to the builder, but 
scheduled builds always go to machine 1, and when pressing the force 
build button also. Seems i need to have a look at the code over the weekend.

Thanks,

Markus


Neil Hemingway wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> That's exactly the config I use, and my builds get distributed across
> my 3 slaves
> 
> Neil
> 
> 
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
> 
> iEYEARECAAYFAkmVdrwACgkQZjaLPuxR95R4lwCfd8llfWXzmP3DRU/T2cwJ944f
> /9YAnRJUVwRSX1DqkDXImCjfyqCt8MV6
> =lYbD
> -----END PGP SIGNATURE-----
> On 13/02/2009, *Markus Kramer* <markus.kramer at mainconcept.de 
> <mailto:markus.kramer at mainconcept.de>> wrote:
> 
>     Hi all,
> 
>     i am just trying to setup some new buildmachines within our buildbot
>     system. The original idea was to have load sharing with some machines,
>     but somehow it does not work as expected.
>     I have setup multiple builders containing the following lines in their
>     build description:
> 
>     'slavenames':[ "win-ia32-xp-b1","win-ia32-xp-b2"],
>     'locks': [xp_sl],
> 
> 
>     In addition i defined the lock like this in the top of config:
> 
>     xp_sl = locks.SlaveLock("cpuxp2",maxCount=1)
> 
> 
>     I would have assumed that due to the fact that SlaveLocks are kept per
>     Slave, this would effectivly distribute the builds among these machines,
>     with only one build going on per Machine, but instead i still see
>     all the builds competing for access to machine b1.
> 
>     Is the availability of locks not considered when selecting buildslave?
>     Is there probably some other method to make each build go to next free
>     slave and only queue when all of them are blocked?
> 
>     Best regards,
> 
>     Markus
> 
> 
> 
> 
>     --
>        Markus Kramer
>        Senior Engineer Codecs
>        Multi-Platform Development Manager
> 
>        MainConcept GmbH
>        Elisabethstr. 1
>        52062 Aachen, Germany
> 
>        Phone: +49 (241) 40108-0
>        Fax:     +49 (241) 40108-10
>        Email:   markus.kramer at mainconcept.com
>     <mailto:markus.kramer at mainconcept.com>
>        Web:    www.mainconcept.com <http://www.mainconcept.com>
> 
>        District Court Aachen: HRB 14788, Managing Directors: Muzaffer
>     Beygirci, David J. Richter, Eric Rodli
> 
> 
>     This message contains confidential, proprietary and/or attorney-client
>     privileged information or material owned by MainConcept and is intended
>     only for the addressed named individual(s).  The information contained
>     in this message may not be copied or distributed without express written
>     consent by an officer of MainConcept.  If the information contained in
>     this email is communicated under a confidentiality or non-disclosure
>     agreement with MainConcept, this email should be considered confidential
>     or proprietary written communication for the purposes of those
>     agreements.  Please notify the sender immediately by e-mail if you have
>     received this e-mail by mistake and delete this e-mail and related
>     information from your system.
> 
> 
>     ------------------------------------------------------------------------------
>     Open Source Business Conference (OSBC), March 24-25, 2009, San
>     Francisco, CA
>     -OSBC tackles the biggest issue in open source: Open Sourcing the
>     Enterprise
>     -Strategies to boost innovation and cut costs with open source
>     participation
>     -Receive a $600 discount off the registration fee with the source
>     code: SFAD
>     http://p.sf.net/sfu/XcvMzF8H
>     _______________________________________________
>     Buildbot-devel mailing list
>     Buildbot-devel at lists.sourceforge.net
>     <mailto:Buildbot-devel at lists.sourceforge.net>
>     https://lists.sourceforge.net/lists/listinfo/buildbot-devel
> 
> 
> 
> 


-- 
   Markus Kramer
   Senior Engineer Codecs
   Multi-Platform Development Manager

   MainConcept GmbH
   Elisabethstr. 1
   52062 Aachen, Germany

   Phone: +49 (241) 40108-0
   Fax:     +49 (241) 40108-10
   Email:   markus.kramer at mainconcept.com
   Web:    www.mainconcept.com

   District Court Aachen: HRB 14788, Managing Directors: Muzaffer 
Beygirci, David J. Richter, Eric Rodli


This message contains confidential, proprietary and/or attorney-client 
privileged information or material owned by MainConcept and is intended 
only for the addressed named individual(s).  The information contained 
in this message may not be copied or distributed without express written 
consent by an officer of MainConcept.  If the information contained in 
this email is communicated under a confidentiality or non-disclosure 
agreement with MainConcept, this email should be considered confidential 
or proprietary written communication for the purposes of those 
agreements.  Please notify the sender immediately by e-mail if you have 
received this e-mail by mistake and delete this e-mail and related 
information from your system.



From macke at yar.nu  Fri Feb 13 14:41:50 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Fri, 13 Feb 2009 15:41:50 +0100
Subject: [Buildbot-devel] SlaveLocks
In-Reply-To: <49958082.609@mainconcept.de>
References: <499555CB.6050409@mainconcept.de>	<47be741a0902130533n7334878fmea1655b9afca8545@mail.gmail.com>
	<49958082.609@mainconcept.de>
Message-ID: <499586AE.6030200@yar.nu>

I think there are a few tickets filed against this. Could to take a look 
at the tickets on load sharing on buildbot.net?

Cheers,
/Marcus

Markus Kramer wrote:
> Hi Neil,
> 
> so is there maybe some difference in versions of buildbot we are using. 
> Mine should all be 0.7.9, but still do not properly share. I can see 
> that i get a connect message for bot slaves to the builder, but 
> scheduled builds always go to machine 1, and when pressing the force 
> build button also. Seems i need to have a look at the code over the weekend.
> 
> Thanks,
> 
> Markus
> 
> 
> Neil Hemingway wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> That's exactly the config I use, and my builds get distributed across
>> my 3 slaves
>>
>> Neil
>>
>>
>> -----BEGIN PGP SIGNATURE-----
>> Version: GnuPG v1.4.9 (GNU/Linux)
>>
>> iEYEARECAAYFAkmVdrwACgkQZjaLPuxR95R4lwCfd8llfWXzmP3DRU/T2cwJ944f
>> /9YAnRJUVwRSX1DqkDXImCjfyqCt8MV6
>> =lYbD
>> -----END PGP SIGNATURE-----
>> On 13/02/2009, *Markus Kramer* <markus.kramer at mainconcept.de 
>> <mailto:markus.kramer at mainconcept.de>> wrote:
>>
>>     Hi all,
>>
>>     i am just trying to setup some new buildmachines within our buildbot
>>     system. The original idea was to have load sharing with some machines,
>>     but somehow it does not work as expected.
>>     I have setup multiple builders containing the following lines in their
>>     build description:
>>
>>     'slavenames':[ "win-ia32-xp-b1","win-ia32-xp-b2"],
>>     'locks': [xp_sl],
>>
>>
>>     In addition i defined the lock like this in the top of config:
>>
>>     xp_sl = locks.SlaveLock("cpuxp2",maxCount=1)
>>
>>
>>     I would have assumed that due to the fact that SlaveLocks are kept per
>>     Slave, this would effectivly distribute the builds among these machines,
>>     with only one build going on per Machine, but instead i still see
>>     all the builds competing for access to machine b1.
>>
>>     Is the availability of locks not considered when selecting buildslave?
>>     Is there probably some other method to make each build go to next free
>>     slave and only queue when all of them are blocked?
>>
>>     Best regards,
>>
>>     Markus
>>
>>
>>
>>
>>     --
>>        Markus Kramer
>>        Senior Engineer Codecs
>>        Multi-Platform Development Manager
>>
>>        MainConcept GmbH
>>        Elisabethstr. 1
>>        52062 Aachen, Germany
>>
>>        Phone: +49 (241) 40108-0
>>        Fax:     +49 (241) 40108-10
>>        Email:   markus.kramer at mainconcept.com
>>     <mailto:markus.kramer at mainconcept.com>
>>        Web:    www.mainconcept.com <http://www.mainconcept.com>
>>
>>        District Court Aachen: HRB 14788, Managing Directors: Muzaffer
>>     Beygirci, David J. Richter, Eric Rodli
>>
>>
>>     This message contains confidential, proprietary and/or attorney-client
>>     privileged information or material owned by MainConcept and is intended
>>     only for the addressed named individual(s).  The information contained
>>     in this message may not be copied or distributed without express written
>>     consent by an officer of MainConcept.  If the information contained in
>>     this email is communicated under a confidentiality or non-disclosure
>>     agreement with MainConcept, this email should be considered confidential
>>     or proprietary written communication for the purposes of those
>>     agreements.  Please notify the sender immediately by e-mail if you have
>>     received this e-mail by mistake and delete this e-mail and related
>>     information from your system.
>>
>>
>>     ------------------------------------------------------------------------------
>>     Open Source Business Conference (OSBC), March 24-25, 2009, San
>>     Francisco, CA
>>     -OSBC tackles the biggest issue in open source: Open Sourcing the
>>     Enterprise
>>     -Strategies to boost innovation and cut costs with open source
>>     participation
>>     -Receive a $600 discount off the registration fee with the source
>>     code: SFAD
>>     http://p.sf.net/sfu/XcvMzF8H
>>     _______________________________________________
>>     Buildbot-devel mailing list
>>     Buildbot-devel at lists.sourceforge.net
>>     <mailto:Buildbot-devel at lists.sourceforge.net>
>>     https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>>
>>
>>
>>
> 
> 




From jean-michel.beuken at uclouvain.be  Fri Feb 13 20:31:42 2009
From: jean-michel.beuken at uclouvain.be (Jean-Michel Beuken)
Date: Fri, 13 Feb 2009 21:31:42 +0100
Subject: [Buildbot-devel] sendchange
Message-ID: <4995D8AE.7000300@uclouvain.be>

Hello,

is it possible to retrieve the values of the arguments of the command

buildbot sendchange --branch='branch' --revision='revision' 
--username='user' ....

in the Builder section in the master.cf  and use WithProperties() to use 
the values ?

for example , I will have needed

WithProperties("bzr+ssh://archives.abinit.org/abinit/5.8/%s/%s","username","branch")

the properties "username"and "branch" don't exist in the context of 
builder at runtime....

thanks in advance...

jmb



From mahatmamanic at gmail.com  Fri Feb 13 22:42:54 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Fri, 13 Feb 2009 14:42:54 -0800
Subject: [Buildbot-devel] Conceptual problem with custom build steps
Message-ID: <c02dbb2a0902131442s34a6ac6ekc63463a5e8d95f0f@mail.gmail.com>

I'm running into trouble pretty quickly with my custom build step, and
having a bit of trouble wrapping my head around what I should be doing with
the code.

I need a build step which will first perform an action in the Buildmaster's
shell, then call a shell command on the Buildslave.  To that end, I wrote a
helper class to take care of the action I wanted to take on the Buildmaster
side, and then wrote my build step like so:

class SpecialBuildStep(ShellCommand):
    def __init__(self, buildMasterCommandObject, dir = None, com = None,
desc = None, descDone = None):
        self._buildMasterCommand = buildMasterCommandObject
        ShellCommand.__init__(self, workdir = dir, command = com,
description = desc, descriptionDone = descDone)

    def start(self):
        self._buildMasterCommandObject.execute()
        ShellCommand.start(self)

This, however, is not working.  I get errors (at build execution time) about
SpecialBuildStep not having a _buildMasterCommand attribute.  If I try to
pass the parameters in for the buildMasterCommand and instantiate it within
the SpecialBuildStep, that causes another kind of problem where (at build
execution time again) I am told that there are missing arguments before my
keyword arguments.

It looks like when I first build the buildsteps and put them in the factory
I am actually building a kind of template which is then later instantiated,
which is fine, but it doesn't seem as though some of my parameters are
surviving the transition.  Debug printing shows the values passed in to
SpecialBuildStep.__init__() to be correct when I am building my factories,
but then seem to disappear later at buil time.

What am I doing wrong?

Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090213/7be0849a/attachment.html>

From mahatmamanic at gmail.com  Fri Feb 13 23:22:14 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Fri, 13 Feb 2009 15:22:14 -0800
Subject: [Buildbot-devel] Conceptual problem with custom build steps
In-Reply-To: <c02dbb2a0902131442s34a6ac6ekc63463a5e8d95f0f@mail.gmail.com>
References: <c02dbb2a0902131442s34a6ac6ekc63463a5e8d95f0f@mail.gmail.com>
Message-ID: <c02dbb2a0902131522l42d2e2cctd773d67fdd91ebd8@mail.gmail.com>

I just realized that I mis-typed the variable name in my example of start()
here.  In reality the name of the variable in the
self._buildMasterCommand.execute() does match the name of the variable in
__init__()

On Fri, Feb 13, 2009 at 2:42 PM, Ian Peters-Campbell <mahatmamanic at gmail.com
> wrote:

> I'm running into trouble pretty quickly with my custom build step, and
> having a bit of trouble wrapping my head around what I should be doing with
> the code.
>
> I need a build step which will first perform an action in the Buildmaster's
> shell, then call a shell command on the Buildslave.  To that end, I wrote a
> helper class to take care of the action I wanted to take on the Buildmaster
> side, and then wrote my build step like so:
>
> class SpecialBuildStep(ShellCommand):
>     def __init__(self, buildMasterCommandObject, dir = None, com = None,
> desc = None, descDone = None):
>         self._buildMasterCommand = buildMasterCommandObject
>         ShellCommand.__init__(self, workdir = dir, command = com,
> description = desc, descriptionDone = descDone)
>
>     def start(self):
>         self._buildMasterCommandObject.execute()
>         ShellCommand.start(self)
>
> This, however, is not working.  I get errors (at build execution time)
> about SpecialBuildStep not having a _buildMasterCommand attribute.  If I try
> to pass the parameters in for the buildMasterCommand and instantiate it
> within the SpecialBuildStep, that causes another kind of problem where (at
> build execution time again) I am told that there are missing arguments
> before my keyword arguments.
>
> It looks like when I first build the buildsteps and put them in the factory
> I am actually building a kind of template which is then later instantiated,
> which is fine, but it doesn't seem as though some of my parameters are
> surviving the transition.  Debug printing shows the values passed in to
> SpecialBuildStep.__init__() to be correct when I am building my factories,
> but then seem to disappear later at buil time.
>
> What am I doing wrong?
>
> Ian
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090213/8d326483/attachment.html>

From macke at yar.nu  Fri Feb 13 14:41:50 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Fri, 13 Feb 2009 15:41:50 +0100
Subject: [Buildbot-devel] SlaveLocks
In-Reply-To: <49958082.609@mainconcept.de>
References: <499555CB.6050409@mainconcept.de>	<47be741a0902130533n7334878fmea1655b9afca8545@mail.gmail.com>
	<49958082.609@mainconcept.de>
Message-ID: <499586AE.6030200@yar.nu>



I think there are a few tickets filed against this. Could to take a look 
at the tickets on load sharing on buildbot.net?

Cheers,
/Marcus

Markus Kramer wrote:
> Hi Neil,
> 
> so is there maybe some difference in versions of buildbot we are using. 
> Mine should all be 0.7.9, but still do not properly share. I can see 
> that i get a connect message for bot slaves to the builder, but 
> scheduled builds always go to machine 1, and when pressing the force 
> build button also. Seems i need to have a look at the code over the weekend.
> 
> Thanks,
> 
> Markus
> 
> 
> Neil Hemingway wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> That's exactly the config I use, and my builds get distributed across
>> my 3 slaves
>>
>> Neil
>>
>>
>> -----BEGIN PGP SIGNATURE-----
>> Version: GnuPG v1.4.9 (GNU/Linux)
>>
>> iEYEARECAAYFAkmVdrwACgkQZjaLPuxR95R4lwCfd8llfWXzmP3DRU/T2cwJ944f
>> /9YAnRJUVwRSX1DqkDXImCjfyqCt8MV6
>> =lYbD
>> -----END PGP SIGNATURE-----
>> On 13/02/2009, *Markus Kramer* <markus.kramer-SLPT3ORPss/5Km1axIVvkA at public.gmane.org 
>> <mailto:markus.kramer-SLPT3ORPss/5Km1axIVvkA at public.gmane.org>> wrote:
>>
>>     Hi all,
>>
>>     i am just trying to setup some new buildmachines within our buildbot
>>     system. The original idea was to have load sharing with some machines,
>>     but somehow it does not work as expected.
>>     I have setup multiple builders containing the following lines in their
>>     build description:
>>
>>     'slavenames':[ "win-ia32-xp-b1","win-ia32-xp-b2"],
>>     'locks': [xp_sl],
>>
>>
>>     In addition i defined the lock like this in the top of config:
>>
>>     xp_sl = locks.SlaveLock("cpuxp2",maxCount=1)
>>
>>
>>     I would have assumed that due to the fact that SlaveLocks are kept per
>>     Slave, this would effectivly distribute the builds among these machines,
>>     with only one build going on per Machine, but instead i still see
>>     all the builds competing for access to machine b1.
>>
>>     Is the availability of locks not considered when selecting buildslave?
>>     Is there probably some other method to make each build go to next free
>>     slave and only queue when all of them are blocked?
>>
>>     Best regards,
>>
>>     Markus
>>
>>
>>
>>
>>     --
>>        Markus Kramer
>>        Senior Engineer Codecs
>>        Multi-Platform Development Manager
>>
>>        MainConcept GmbH
>>        Elisabethstr. 1
>>        52062 Aachen, Germany
>>
>>        Phone: +49 (241) 40108-0
>>        Fax:     +49 (241) 40108-10
>>        Email:   markus.kramer-SLPT3ORPss9aodhZ+FW2PA at public.gmane.org
>>     <mailto:markus.kramer-SLPT3ORPss9aodhZ+FW2PA at public.gmane.org>
>>        Web:    www.mainconcept.com <http://www.mainconcept.com>
>>
>>        District Court Aachen: HRB 14788, Managing Directors: Muzaffer
>>     Beygirci, David J. Richter, Eric Rodli
>>
>>
>>     This message contains confidential, proprietary and/or attorney-client
>>     privileged information or material owned by MainConcept and is intended
>>     only for the addressed named individual(s).  The information contained
>>     in this message may not be copied or distributed without express written
>>     consent by an officer of MainConcept.  If the information contained in
>>     this email is communicated under a confidentiality or non-disclosure
>>     agreement with MainConcept, this email should be considered confidential
>>     or proprietary written communication for the purposes of those
>>     agreements.  Please notify the sender immediately by e-mail if you have
>>     received this e-mail by mistake and delete this e-mail and related
>>     information from your system.
>>
>>
>>     ------------------------------------------------------------------------------
>>     Open Source Business Conference (OSBC), March 24-25, 2009, San
>>     Francisco, CA
>>     -OSBC tackles the biggest issue in open source: Open Sourcing the
>>     Enterprise
>>     -Strategies to boost innovation and cut costs with open source
>>     participation
>>     -Receive a $600 discount off the registration fee with the source
>>     code: SFAD
>>     http://p.sf.net/sfu/XcvMzF8H
>>     _______________________________________________
>>     Buildbot-devel mailing list
>>     Buildbot-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f at public.gmane.org
>>     <mailto:Buildbot-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f at public.gmane.org>
>>     https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>>
>>
>>
>>
> 
> 





From dustin at zmanda.com  Sun Feb 15 23:40:05 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Sun, 15 Feb 2009 18:40:05 -0500
Subject: [Buildbot-devel] Conceptual problem with custom build steps
In-Reply-To: <c02dbb2a0902131442s34a6ac6ekc63463a5e8d95f0f@mail.gmail.com>
References: <c02dbb2a0902131442s34a6ac6ekc63463a5e8d95f0f@mail.gmail.com>
Message-ID: <42338fbf0902151540g3d788100j294f71aefbee4025@mail.gmail.com>

On Fri, Feb 13, 2009 at 5:42 PM, Ian Peters-Campbell
<mahatmamanic at gmail.com> wrote:
> This, however, is not working.  I get errors (at build execution time) about
> SpecialBuildStep not having a _buildMasterCommand attribute.  If I try to
> pass the parameters in for the buildMasterCommand and instantiate it within
> the SpecialBuildStep, that causes another kind of problem where (at build
> execution time again) I am told that there are missing arguments before my
> keyword arguments.

This is a problem with arrangement where BuildStep objects are their
own factories.

This is how my constructor looks:

 17     def __init__(self, command, **kwargs):
 18         BuildStep.__init__(self, **kwargs)
 19         self.addFactoryArguments(command=command)
 20         self.command=command

On a larger note, I have a MasterShellCommand class that I'll get
committed shortly, which may do what you need.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Mon Feb 16 01:07:31 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Sun, 15 Feb 2009 20:07:31 -0500
Subject: [Buildbot-devel] Conceptual problem with custom build steps
In-Reply-To: <42338fbf0902151540g3d788100j294f71aefbee4025@mail.gmail.com>
References: <c02dbb2a0902131442s34a6ac6ekc63463a5e8d95f0f@mail.gmail.com>
	<42338fbf0902151540g3d788100j294f71aefbee4025@mail.gmail.com>
Message-ID: <42338fbf0902151707pc92af12j79e0f08495874050@mail.gmail.com>

On Sun, Feb 15, 2009 at 6:40 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
> On a larger note, I have a MasterShellCommand class that I'll get
> committed shortly, which may do what you need.

Commit is here:

http://github.com/djmitche/buildbot/commit/77b386426c2ee498e43185553b756224db74fea8

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From mahatmamanic at gmail.com  Mon Feb 16 04:33:43 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Sun, 15 Feb 2009 20:33:43 -0800
Subject: [Buildbot-devel] Conceptual problem with custom build steps
In-Reply-To: <42338fbf0902151707pc92af12j79e0f08495874050@mail.gmail.com>
References: <c02dbb2a0902131442s34a6ac6ekc63463a5e8d95f0f@mail.gmail.com>
	<42338fbf0902151540g3d788100j294f71aefbee4025@mail.gmail.com>
	<42338fbf0902151707pc92af12j79e0f08495874050@mail.gmail.com>
Message-ID: <c02dbb2a0902152033n5f971092he12db17c339ac83a@mail.gmail.com>

Good on ya Dustin, MasterShellCommand looks like exactly what I need.  Also
thanks for the explanation, the self-factory is definitely where the
disconnect was.  I assume if I was inheriting from ShellCommand I would
still want to upcall to its __init__() rather than to BuildStep's right?
Would I also need to do any stripping from the kwargs dict the way
ShellCommand seems to before it upcalls to its parent, or is that specific
to ShellCommand?

Ian



On Sun, Feb 15, 2009 at 5:07 PM, Dustin J. Mitchell <dustin at zmanda.com>wrote:

> On Sun, Feb 15, 2009 at 6:40 PM, Dustin J. Mitchell <dustin at zmanda.com>
> wrote:
> > On a larger note, I have a MasterShellCommand class that I'll get
> > committed shortly, which may do what you need.
>
> Commit is here:
>
>
> http://github.com/djmitche/buildbot/commit/77b386426c2ee498e43185553b756224db74fea8
>
> Dustin
>
> --
> Storage Software Engineer
> http://www.zmanda.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090215/c6b1890f/attachment.html>

From dustin at zmanda.com  Mon Feb 16 04:40:52 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Sun, 15 Feb 2009 23:40:52 -0500
Subject: [Buildbot-devel] Conceptual problem with custom build steps
In-Reply-To: <c02dbb2a0902152033n5f971092he12db17c339ac83a@mail.gmail.com>
References: <c02dbb2a0902131442s34a6ac6ekc63463a5e8d95f0f@mail.gmail.com>
	<42338fbf0902151540g3d788100j294f71aefbee4025@mail.gmail.com>
	<42338fbf0902151707pc92af12j79e0f08495874050@mail.gmail.com>
	<c02dbb2a0902152033n5f971092he12db17c339ac83a@mail.gmail.com>
Message-ID: <42338fbf0902152040j25e08364x71f3bddd1040d03f@mail.gmail.com>

On Sun, Feb 15, 2009 at 11:33 PM, Ian Peters-Campbell
<mahatmamanic at gmail.com> wrote:
> I assume if I was inheriting from ShellCommand I would still want to upcall
> to its __init__() rather than to BuildStep's right?

Yep.

> Would I also need to do
> any stripping from the kwargs dict the way ShellCommand seems to before it
> upcalls to its parent, or is that specific to ShellCommand?

I *think* that's specific to ShellCommand.  But this isn't a
particularly strong area of Python, unfortunately.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Mon Feb 16 05:25:40 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Mon, 16 Feb 2009 00:25:40 -0500
Subject: [Buildbot-devel] 0.7.10 release update
Message-ID: <42338fbf0902152125o42cfc414x7da344cb7ff3a4ca@mail.gmail.com>

Well, I just finished triaging all of the "undecided" bugs, leaving 33
unfinished bugs in the 0.7.10 milestone:
  http://buildbot.net/trac/query?status=new&status=assigned&status=reopened&milestone=0.7.10

I'd like to get a release candidate for 0.7.10 out in about a week.
I've assigned a number of these bugs to myself or others, but there
are 15 unclaimed bugs:
  http://buildbot.net/trac/query?status=new&status=assigned&status=reopened&milestone=0.7.10&order=priority&owner=

Most of these are probably 30m worth of work, so please make a point
of looking over this list, picking a bug that looks fixable, and
hacking on it sometime before next Sunday.

Thanks for hacking on buildbot!

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From gerg.ward+buildbot at gmail.com  Mon Feb 16 19:35:41 2009
From: gerg.ward+buildbot at gmail.com (Greg Ward)
Date: Mon, 16 Feb 2009 14:35:41 -0500
Subject: [Buildbot-devel] Dynamic list of logfiles
Message-ID: <20090216193541.GA14279@intelerad.com>

Hi all --

I have a ShellCommand that creates log files based on what it finds in
its build tree.  Specifically, it's a wrapper around "rpmbuild" that
processes <dir>/*.spec, creating one log for each .spec file
processed.

Currently, I have cobbled something together where I use a SetProperty
step to look for the same list of *.spec files and derive the log
filenames from that.  That's a bit gross because I'm repeating two
bits of logic (find *.spec and transmute foo.spec to foo.log) in two
places (the script being run and master.cfg).

It has occurred to me that a ShellCommand that dynamically discovers
log files would be really cool.  I think I want something like this:

  ShellCommand(name="build-rpms",
               command=["./buildrpms", 
                        "--spec-dir", "spec", 
                        "--log-dir", "log"],
               logfilepattern="log/buildrpms.*.log")

Then, as my buildrpms script is running, SlaveShellCommand (I think)
would periodically look for files matching my logfilepattern and
update the step's log files accordingly.

Does this sound crazy?  Has anyone attempted this?  Is it possible
to do entirely within master.cfg, or would I need to patch slave-side
code?

Thanks --

       Greg



From bbb at illinois.edu  Mon Feb 16 20:08:04 2009
From: bbb at illinois.edu (Bill Baker)
Date: Mon, 16 Feb 2009 14:08:04 -0600
Subject: [Buildbot-devel] Dynamic list of logfiles
In-Reply-To: <20090216193541.GA14279@intelerad.com>
References: <20090216193541.GA14279@intelerad.com>
Message-ID: <8016c6df0902161208q598107c5g72c6e7c84dd3d85c@mail.gmail.com>

I have half of that -- the slave automatically posts files to an external
web server, and the BuildStep notices in stdio and adds a link.  Straight
from master.cfg:

class LogFileShellCommand(ShellCommand):
    log_files = []
    log_files.append("config.log")
    log_files.append("build.log")

    def __init__(self, **kwargs):
        # And upcall to let the base class do its work
        ShellCommand.__init__(self, **kwargs)

    def createSummary(self, log):
        for line in log.getText().split("\n"):
            for log_file in self.log_files:
                # line format is "    log file build.log saved to http://
..."
                if (line.startswith("log file ") and line.find(" saved to ")
>= 0):
                    split = line.split(" saved to ", 1)
                    self.addURL(split[0].split("log file ")[1], split[1])

It's barely functional enough to work -- you can probably see easy ways to
make it more systematic; I think the tricky part is saving the log files --
that requires some logic on the slave side, one way or another.

On Mon, Feb 16, 2009 at 1:35 PM, Greg Ward
<gerg.ward+buildbot at gmail.com<gerg.ward%2Bbuildbot at gmail.com>
> wrote:

> Hi all --
>
> I have a ShellCommand that creates log files based on what it finds in
> its build tree.  Specifically, it's a wrapper around "rpmbuild" that
> processes <dir>/*.spec, creating one log for each .spec file
> processed.
>
> Currently, I have cobbled something together where I use a SetProperty
> step to look for the same list of *.spec files and derive the log
> filenames from that.  That's a bit gross because I'm repeating two
> bits of logic (find *.spec and transmute foo.spec to foo.log) in two
> places (the script being run and master.cfg).
>
> It has occurred to me that a ShellCommand that dynamically discovers
> log files would be really cool.  I think I want something like this:
>
>  ShellCommand(name="build-rpms",
>               command=["./buildrpms",
>                        "--spec-dir", "spec",
>                        "--log-dir", "log"],
>               logfilepattern="log/buildrpms.*.log")
>
> Then, as my buildrpms script is running, SlaveShellCommand (I think)
> would periodically look for files matching my logfilepattern and
> update the step's log files accordingly.
>
> Does this sound crazy?  Has anyone attempted this?  Is it possible
> to do entirely within master.cfg, or would I need to patch slave-side
> code?
>
> Thanks --
>
>       Greg
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
> CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the
> Enterprise
> -Strategies to boost innovation and cut costs with open source
> participation
> -Receive a $600 discount off the registration fee with the source code:
> SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090216/3e2ec267/attachment.html>

From smilner at redhat.com  Mon Feb 16 20:35:43 2009
From: smilner at redhat.com (Steve 'Ashcrow' Milner)
Date: Mon, 16 Feb 2009 15:35:43 -0500
Subject: [Buildbot-devel] OT [Re: Dynamic list of logfiles]
In-Reply-To: <20090216193541.GA14279@intelerad.com>
References: <20090216193541.GA14279@intelerad.com>
Message-ID: <20090216203543.GI2885@arc.rdu.redhat.com>

On 16/02/09 14:35 -0500, Greg Ward wrote:
>Hi all --
>
>I have a ShellCommand that creates log files based on what it finds in
>its build tree.  Specifically, it's a wrapper around "rpmbuild" that
>processes <dir>/*.spec, creating one log for each .spec file
>processed.
>
>Currently, I have cobbled something together where I use a SetProperty
>step to look for the same list of *.spec files and derive the log
>filenames from that.  That's a bit gross because I'm repeating two
>bits of logic (find *.spec and transmute foo.spec to foo.log) in two
>places (the script being run and master.cfg).
>
>It has occurred to me that a ShellCommand that dynamically discovers
>log files would be really cool.  I think I want something like this:
>
>  ShellCommand(name="build-rpms",
>               command=["./buildrpms", 
>                        "--spec-dir", "spec", 
>                        "--log-dir", "log"],
>               logfilepattern="log/buildrpms.*.log")
>
>Then, as my buildrpms script is running, SlaveShellCommand (I think)
>would periodically look for files matching my logfilepattern and
>update the step's log files accordingly.
>
>Does this sound crazy?  Has anyone attempted this?  Is it possible
>to do entirely within master.cfg, or would I need to patch slave-side
>code?
>
>Thanks --
>
>       Greg
>
>------------------------------------------------------------------------------
>Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
>-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
>-Strategies to boost innovation and cut costs with open source participation
>-Receive a $600 discount off the registration fee with the source code: SFAD
>http://p.sf.net/sfu/XcvMzF8H
>_______________________________________________
>Buildbot-devel mailing list
>Buildbot-devel at lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/buildbot-devel

On a side note, check out the rpm code comming down in a future
version
http://github.com/djmitche/buildbot/tree/aa3a5bc6ee13859022b4ad13283b34f7ccf78781/buildbot/steps/package/rpm.

-- 
kthxbye!
Steve 'Ashcrow' Milner
Agent of Infosec
IRC: ashcrow
GnuPG ID: 28DFD4BE

"In the heat of conversation I may have said certain things I believe 
to be untrue. The alleged lie that you might have heard me saying 
allegedly moments ago ... thats a parasite that lives in my neck." 
     -- Tad Ghostal
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://buildbot.net/pipermail/devel/attachments/20090216/d9e90a64/attachment.bin>

From Y.Itskovich at Astronautics.com  Mon Feb 16 22:37:02 2009
From: Y.Itskovich at Astronautics.com (Yury Itskovich)
Date: Mon, 16 Feb 2009 16:37:02 -0600
Subject: [Buildbot-devel] why does 'try' apply patch to triggeredl build?
Message-ID: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DE4@HQMAIL1.aca.astronautics.com>

I have a Try_Userpass scheduler that builds my app, and a Triggerable scheduler (with a separate builder/factory) that builds and runs my test app (the test app build depends on the primary app build).  The problem is, Buildbot tries to apply the primary app patch to the test app build as well -- why?  Is there a simple way to prevent this?  Maybe I could override this behavior with a subclass or something, unless I'm missing something obvious.

Thanks,
Yury
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090216/338fc0c3/attachment.html>

From l10n.moz at googlemail.com  Mon Feb 16 22:53:22 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Mon, 16 Feb 2009 23:53:22 +0100
Subject: [Buildbot-devel] Dynamic list of logfiles
In-Reply-To: <20090216193541.GA14279@intelerad.com>
References: <20090216193541.GA14279@intelerad.com>
Message-ID: <9de1bdb60902161453q2b30f097s3aad085ba14def38@mail.gmail.com>

2009/2/16 Greg Ward
<gerg.ward+buildbot at gmail.com<gerg.ward%2Bbuildbot at gmail.com>
>

> Hi all --
>
> I have a ShellCommand that creates log files based on what it finds in
> its build tree.  Specifically, it's a wrapper around "rpmbuild" that
> processes <dir>/*.spec, creating one log for each .spec file
> processed.
>
> Currently, I have cobbled something together where I use a SetProperty
> step to look for the same list of *.spec files and derive the log
> filenames from that.  That's a bit gross because I'm repeating two
> bits of logic (find *.spec and transmute foo.spec to foo.log) in two
> places (the script being run and master.cfg).
>
> It has occurred to me that a ShellCommand that dynamically discovers
> log files would be really cool.  I think I want something like this:
>
>  ShellCommand(name="build-rpms",
>               command=["./buildrpms",
>                        "--spec-dir", "spec",
>                        "--log-dir", "log"],
>               logfilepattern="log/buildrpms.*.log")
>
> Then, as my buildrpms script is running, SlaveShellCommand (I think)
> would periodically look for files matching my logfilepattern and
> update the step's log files accordingly.
>
> Does this sound crazy?  Has anyone attempted this?  Is it possible
> to do entirely within master.cfg, or would I need to patch slave-side
> code?
>

I'm not sure if logfilepattern is really that much of a use case in general
that it'd be worth adding that in general.

Here's what you had to do in principle:

After the step finishes, find all logs, transfer them to the master,
addCompleteLog there, or do a addLog().... They would then magically show up
in the web interface, too.

I would think that you had to do this when the step is actually done, as
you'd otherwise had to not just transfer the files, but watch for changes,
and update the upstream content. In general, you probably can't ensure that
those files are only appended to, so you'd run into issues when not.

All in all nothing too scary, but yes, you would need to hack python code on
the slave, and make sure that all your slaves get your custom buildbot, or
your buildbot customization.

Axel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090216/584ad510/attachment.html>

From Y.Itskovich at Astronautics.com  Mon Feb 16 22:39:28 2009
From: Y.Itskovich at Astronautics.com (Yury Itskovich)
Date: Mon, 16 Feb 2009 16:39:28 -0600
Subject: [Buildbot-devel] need help using "try"
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDB@HQMAIL1.aca.astronautics.com>	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDC@HQMAIL1.aca.astronautics.com>	<4987FDFE.9090603@yar.nu>	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDD@HQMAIL1.aca.astronautics.com>
	<79D0CFE134D11C45A1E70E6EF187EBFC01EF1DDE@HQMAIL1.aca.astronautics.com>
	<49894C39.1030309@yar.nu>
Message-ID: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DE5@HQMAIL1.aca.astronautics.com>

Forgot to reply: of course, the slave wasn't running...)
Thanks again.

Yury


-----Original Message-----
From: Marcus Lindblom [mailto:macke at yar.nu]
Sent: Wed 2/4/2009 2:05 AM
To: Yury Itskovich
Cc: Marcus Lindblom; public-buildbot-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f at hugh.gmane.org
Subject: Re: need help using "try"
 


Hi Yury,

Yury Itskovich wrote:

> 2009-02-03 09:01:36-0600 [Broker,4,127.0.0.1] <Builder 'buildbot-full' 
> at 144934220>: want to start build, but we don't have a remote

[snip]

> 2009-02-02 14:36:11-0600 [-] maybeStartBuild <Builder 'buildbot-full' at 
> 144937836>: [<buildbot.process.base.BuildRequest instance at 0x8a5972c>] []
> 2009-02-02 14:36:11-0600 [-] <Builder 'buildbot-full' at 144937836>: 
> want to start build, but we don't have a remote
> 
> I must be missing something simple in the setup/configuration that's 
> causing this.

'No remote' might mean no buildslave? Is the 'testslave' up and running 
properly.

I haven't seen that error message though, so I'm guesstimating. :-/

Cheers,
/Marcus




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090216/b6751d21/attachment.html>

From Y.Itskovich at Astronautics.com  Tue Feb 17 00:12:23 2009
From: Y.Itskovich at Astronautics.com (Yury Itskovich)
Date: Mon, 16 Feb 2009 18:12:23 -0600
Subject: [Buildbot-devel] why does 'try' apply patch to triggeredl build?
References: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DE4@HQMAIL1.aca.astronautics.com>
Message-ID: <79D0CFE134D11C45A1E70E6EF187EBFC01EF1DE7@HQMAIL1.aca.astronautics.com>

On second thought, I don't really need to do it this way: the test app code doesn't actually depend on the primary app code...  Still, the behavior doesn't seem very intuitive.


-----Original Message-----
From: Yury Itskovich
Sent: Mon 2/16/2009 4:37 PM
To: buildbot-devel at lists.sourceforge.net
Subject: why does 'try' apply patch to triggeredl build?
 
I have a Try_Userpass scheduler that builds my app, and a Triggerable scheduler (with a separate builder/factory) that builds and runs my test app (the test app build depends on the primary app build).  The problem is, Buildbot tries to apply the primary app patch to the test app build as well -- why?  Is there a simple way to prevent this?  Maybe I could override this behavior with a subclass or something, unless I'm missing something obvious.

Thanks,
Yury

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090216/1ad4b79c/attachment.html>

From gerg.ward+buildbot at gmail.com  Tue Feb 17 01:25:31 2009
From: gerg.ward+buildbot at gmail.com (Greg Ward)
Date: Mon, 16 Feb 2009 20:25:31 -0500
Subject: [Buildbot-devel] Dynamic list of logfiles
In-Reply-To: <8016c6df0902161208q598107c5g72c6e7c84dd3d85c@mail.gmail.com>
References: <20090216193541.GA14279@intelerad.com>
	<8016c6df0902161208q598107c5g72c6e7c84dd3d85c@mail.gmail.com>
Message-ID: <20090217012531.GA29529@intelerad.com>

On 16 February 2009, Bill Baker said:
> I have half of that -- the slave automatically posts files to an external
> web server, and the BuildStep notices in stdio and adds a link.  Straight
> from master.cfg:
> 
> class LogFileShellCommand(ShellCommand):
>     log_files = []
>     log_files.append("config.log")
>     log_files.append("build.log")
> 
>     def __init__(self, **kwargs):
>         # And upcall to let the base class do its work
>         ShellCommand.__init__(self, **kwargs)
> 
>     def createSummary(self, log):
>         for line in log.getText().split("\n"):
>             for log_file in self.log_files:
>                 # line format is "    log file build.log saved to http://..."
>                 if (line.startswith("log file ") and line.find(" saved to ") >= 0):
>                     split = line.split(" saved to ", 1)
>                     self.addURL(split[0].split("log file ")[1], split[1])

What a charming hack!  As long as the slave command cooperates and
prints something like "logging to foo.log", you can take advantage of
that and suss out the log file URLs or filenames.  (BTW, you could
make your code shorter and arguably clearer with a regex there.)

You inspired me to try something similar but slightly more powerful:
  * use a LogLineObserver to watch for those "logging to foo.log" lines
    in my slave command's stdout
  * when I see one, turn it into a LogFile and start watching it

If this had worked, it would have been super cool and would have done
everything I wanted.  Alas!  This appears to be the first nifty hack
I've tried with Buildbot that falls down flat, and it's because the
slave-side code doesn't let you add new log files while a command is
running.  Specifically, buildbot.slave.command.ShellCommand's
attribute logFileWatchers list is populated in the constructor and not
changed again.  So there is no way for the master to tell the slave
"start watching foo.log now".  Darn.

I suspect that dynamic log file nirvana requires a protocol change.
Sod it.  I'll see if I can figure out a way to do this after the step
completes.

       Greg



From macke at yar.nu  Tue Feb 17 08:18:22 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 17 Feb 2009 09:18:22 +0100
Subject: [Buildbot-devel] buildbot unable to kill windows process after
 timeout...
In-Reply-To: <393953.54146.qm@web50006.mail.re2.yahoo.com>
References: <1833.95094.qm@web50008.mail.re2.yahoo.com>
	<490F2A18.3050603@yar.nu>
	<393953.54146.qm@web50006.mail.re2.yahoo.com>
Message-ID: <499A72CE.90807@yar.nu>

(fixed top posting)

> Jimmy John wrote:
>> From: Marcus Lindblom <macke at yar.nu>
>>
>> There is a twisted patch that might help with this. (JobControl on 
>> Win32). Search twisted/buidbot trac or the buildbot mail-list.
>>     
> Has anybody had any luck with running a buildbot slave on windows 2003 server?
>
> I tried the patch at http://twistedmatrix.com/trac/ticket/2726 but it does not work. If the slave times out then the master is unable to kill it and does not proceed further. eg.
>
> command timed out: 60 seconds without output, killing pid 1548
> SIGKILL failed to kill process
> using fake rc=-1
> program finished with exit code -1
>
> remoteFailed: [Failure instance: Traceback from remote host -- Traceback (most recent call last):
> Failure: buildbot.slave.commands.TimeoutError: SIGKILL failed to kill process
>
> ]
>   
Hm. I haven't tried 2003 server for build slaves, only XP Pro. (The 
patch worked there.) Maybe someone else on the list has an idea??

I'm just guessing here, but perhaps the policies/rules are different on 
2003 Server compared to XP about processes stopping another processes?

Cheers,
/Marcus



From dustin at zmanda.com  Tue Feb 17 16:24:56 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Tue, 17 Feb 2009 11:24:56 -0500
Subject: [Buildbot-devel] PyCon sprint?
Message-ID: <42338fbf0902170824r464239c4q32362316c5d51156@mail.gmail.com>

Reed O'Brien has suggested that we have a buildbot-1.0 sprint sometime
during/around PyCon, probably in the range of 3/30-4/2.  I'm not
planning to attend PyCon, but I lived in Chicago long enough that I
can easily find somewhere to crash if others will be in town and want
to sprint.

So, if you're interested, and will be in Chicago around those dates,
let me know.  I need to see if there's a "critical mass" before
developing the idea further.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From smilner at redhat.com  Tue Feb 17 18:30:13 2009
From: smilner at redhat.com (Steve 'Ashcrow' Milner)
Date: Tue, 17 Feb 2009 13:30:13 -0500
Subject: [Buildbot-devel] PyCon sprint?
In-Reply-To: <42338fbf0902170824r464239c4q32362316c5d51156@mail.gmail.com>
References: <42338fbf0902170824r464239c4q32362316c5d51156@mail.gmail.com>
Message-ID: <20090217183013.GA4514@arc.rdu.redhat.com>

On 17/02/09 11:24 -0500, Dustin J. Mitchell wrote:
>Reed O'Brien has suggested that we have a buildbot-1.0 sprint sometime
>during/around PyCon, probably in the range of 3/30-4/2.  I'm not
>planning to attend PyCon, but I lived in Chicago long enough that I
>can easily find somewhere to crash if others will be in town and want
>to sprint.
>
>So, if you're interested, and will be in Chicago around those dates,
>let me know.  I need to see if there's a "critical mass" before
>developing the idea further.
>
>Dustin
>
>-- 
>Storage Software Engineer
>http://www.zmanda.com
>
>------------------------------------------------------------------------------
>Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
>-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
>-Strategies to boost innovation and cut costs with open source participation
>-Receive a $600 discount off the registration fee with the source code: SFAD
>http://p.sf.net/sfu/XcvMzF8H
>_______________________________________________
>Buildbot-devel mailing list
>Buildbot-devel at lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/buildbot-devel

I'm not in Chicago nor can I make it to pycon this year BUT if there
is an online portion to the sprint it would open it up to folks that
may not be able to attend otherwise.

-- 
kthxbye!
Steve 'Ashcrow' Milner
Agent of Infosec
IRC: ashcrow
GnuPG ID: 28DFD4BE

"In the heat of conversation I may have said certain things I believe 
to be untrue. The alleged lie that you might have heard me saying 
allegedly moments ago ... thats a parasite that lives in my neck." 
     -- Tad Ghostal
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://buildbot.net/pipermail/devel/attachments/20090217/09e3f9fd/attachment.bin>

From dustin at zmanda.com  Tue Feb 17 19:32:44 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Tue, 17 Feb 2009 14:32:44 -0500
Subject: [Buildbot-devel] PyCon sprint?
In-Reply-To: <20090217183013.GA4514@arc.rdu.redhat.com>
References: <42338fbf0902170824r464239c4q32362316c5d51156@mail.gmail.com>
	<20090217183013.GA4514@arc.rdu.redhat.com>
Message-ID: <42338fbf0902171132p7a713fc2g2322d257516f6aa@mail.gmail.com>

On Tue, Feb 17, 2009 at 1:30 PM, Steve 'Ashcrow' Milner
<smilner at redhat.com> wrote:
> I'm not in Chicago nor can I make it to pycon this year BUT if there
> is an online portion to the sprint it would open it up to folks that
> may not be able to attend otherwise.

There will definitely be such a component, if it happens.  At the
moment, what I'm trying to ascertain is whether it's worth the expense
to fly to Chicago.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From mahatmamanic at gmail.com  Tue Feb 17 21:34:24 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Tue, 17 Feb 2009 13:34:24 -0800
Subject: [Buildbot-devel] build notification/change logs
Message-ID: <c02dbb2a0902171334j7fa0ac32t66a08c500d028313@mail.gmail.com>

Hey all,

My build system has two kinds of builds: continuous integration builds,
which are automatically performed and only send out email to developers, and
only if the build fails, and then release builds, which are specifically
ordered by a developer, are packaged, distributed, and email is sent to the
entire project team including dev, QA, product management, etc.

What I would like is to include SVN commit notes with the outgoing email.  I
would like for the email to include all the SVN notes that are more recent
than the last release build (so the email explains the differential from the
last release build.)  Is there an automagic way to accomplish this, or
should I put together a little internal data structure that I pickle out
between uses and get revision numbers from for grabbing change notes?

Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090217/fd91e8f2/attachment.html>

From dustin at zmanda.com  Tue Feb 17 23:50:31 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Tue, 17 Feb 2009 18:50:31 -0500
Subject: [Buildbot-devel] build notification/change logs
In-Reply-To: <c02dbb2a0902171334j7fa0ac32t66a08c500d028313@mail.gmail.com>
References: <c02dbb2a0902171334j7fa0ac32t66a08c500d028313@mail.gmail.com>
Message-ID: <42338fbf0902171550y793b881er5f7cde474061e343@mail.gmail.com>

On Tue, Feb 17, 2009 at 4:34 PM, Ian Peters-Campbell
<mahatmamanic at gmail.com> wrote:
> What I would like is to include SVN commit notes with the outgoing email.  I
> would like for the email to include all the SVN notes that are more recent
> than the last release build (so the email explains the differential from the
> last release build.)  Is there an automagic way to accomplish this, or
> should I put together a little internal data structure that I pickle out
> between uses and get revision numbers from for grabbing change notes?

I would recommend putting this in a script in your source repository,
and adding a ShellCommand to your release build that invokes the
script.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Wed Feb 18 22:41:59 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 18 Feb 2009 17:41:59 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <77fcfd0a0902111011i61f91402h6ef589d89f113b0e@mail.gmail.com>
References: <49930F38.2040400@starswan.com>
	<42338fbf0902111001w2195bfc7o6c8846d65f65cb5a@mail.gmail.com>
	<77fcfd0a0902111011i61f91402h6ef589d89f113b0e@mail.gmail.com>
Message-ID: <42338fbf0902181441h61a2b6b6s37c57ddab066d44d@mail.gmail.com>

What's the status of this proposal?  I'd really like to get this set
up, if possible in time to find and fix some bugs before the 0.7.10
release!

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Thu Feb 19 03:58:33 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 18 Feb 2009 22:58:33 -0500
Subject: [Buildbot-devel] PyCon sprint?
In-Reply-To: <42338fbf0902170824r464239c4q32362316c5d51156@mail.gmail.com>
References: <42338fbf0902170824r464239c4q32362316c5d51156@mail.gmail.com>
Message-ID: <42338fbf0902181958x5c6336c5y2b124c428270fe48@mail.gmail.com>

On Tue, Feb 17, 2009 at 11:24 AM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
> Reed O'Brien has suggested that we have a buildbot-1.0 sprint sometime
> during/around PyCon, probably in the range of 3/30-4/2.  I'm not
> planning to attend PyCon, but I lived in Chicago long enough that I
> can easily find somewhere to crash if others will be in town and want
> to sprint.

At this point, I definitely will be in town on those days, but since I
don't hear anyone excited about a sprint, I won't be attending PyCon
or hosting a sprint.  If interested folks turn out to be in town, we
can certainly meet up somewhere (maybe actually *in* Chicago, instead
of at O'Hare!) to get some hacking done.  Let me know.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From stephen at starswan.com  Thu Feb 19 20:29:36 2009
From: stephen at starswan.com (Stephen Dicks)
Date: Thu, 19 Feb 2009 20:29:36 +0000
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <42338fbf0902190725w4566d6a2sfb0da3ea641a68eb@mail.gmail.com>
References: <112350350473f2dba311.2089018456.stephen@starswan.com>
	<42338fbf0902190725w4566d6a2sfb0da3ea641a68eb@mail.gmail.com>
Message-ID: <499DC130.70307@starswan.com>

Dustin J. Mitchell wrote:
> On Thu, Feb 19, 2009 at 4:17 AM, Stephen Dicks <stephen at starswan.com> wrote:
>   
>> I'm currently looking at setting it up on my telehouse-hosted debian box, but I came across some installation issues this morning & I'm pretty much tied up today. Might get a chance to look at it later or tomorrow.
>>     
>
> Great -- good to know it's still in progress.  Is there anything I or
> other list members can do to help?  Maybe we can start drumming up
> donated buildslaves?
>
>   
Yes - a MacOSX slave would be very useful, as I don't currently own any 
Mac hardware.

--
Steve



From rmecklenburg at control4.com  Thu Feb 19 20:44:13 2009
From: rmecklenburg at control4.com (Robert Mecklenburg)
Date: Thu, 19 Feb 2009 13:44:13 -0700
Subject: [Buildbot-devel] downloading email archives from sourceforge
Message-ID: <18845.50333.993038.921661@gargle.gargle.HOWL>

This is a lame post and I apologize in advance, but...

Is there any way to download the email archives for the last year as a
set of tar.gz files?  I find the sourceforge mail archive browser to
be extremely tedious.  Unfortunately, I have not been able to discover
where (or if) these archive files are available.

Thanks,
-- 
Robert



From dustin at zmanda.com  Thu Feb 19 21:01:31 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Thu, 19 Feb 2009 16:01:31 -0500
Subject: [Buildbot-devel] downloading email archives from sourceforge
In-Reply-To: <18845.50333.993038.921661@gargle.gargle.HOWL>
References: <18845.50333.993038.921661@gargle.gargle.HOWL>
Message-ID: <42338fbf0902191301i2a31c3d1h512696b8fc7c396@mail.gmail.com>

On Thu, Feb 19, 2009 at 3:44 PM, Robert Mecklenburg
<rmecklenburg at control4.com> wrote:
> Is there any way to download the email archives for the last year as a
> set of tar.gz files?  I find the sourceforge mail archive browser to
> be extremely tedious.  Unfortunately, I have not been able to discover
> where (or if) these archive files are available.

To my knowledge, that's not available to us as project admins.  I
agree that the browser (not to put words in your mouth) sucks.  There
are a few other archives around, though; my favorite is
  http://news.gmane.org/gmane.comp.python.buildbot.devel

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From bhearsum at mozilla.com  Thu Feb 19 21:04:55 2009
From: bhearsum at mozilla.com (Ben Hearsum)
Date: Thu, 19 Feb 2009 16:04:55 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <499DC130.70307@starswan.com>
References: <112350350473f2dba311.2089018456.stephen@starswan.com>
	<42338fbf0902190725w4566d6a2sfb0da3ea641a68eb@mail.gmail.com>
	<499DC130.70307@starswan.com>
Message-ID: <5E44F324-3AFE-440B-A009-63A9122ED5D0@mozilla.com>

I might be able to get a mac mini or two for this purpose.
On 19-Feb-09, at 3:29 PM, Stephen Dicks wrote:

> Dustin J. Mitchell wrote:
>> On Thu, Feb 19, 2009 at 4:17 AM, Stephen Dicks  
>> <stephen at starswan.com> wrote:
>>
>>> I'm currently looking at setting it up on my telehouse-hosted  
>>> debian box, but I came across some installation issues this  
>>> morning & I'm pretty much tied up today. Might get a chance to  
>>> look at it later or tomorrow.
>>>
>>
>> Great -- good to know it's still in progress.  Is there anything I or
>> other list members can do to help?  Maybe we can start drumming up
>> donated buildslaves?
>>
>>
> Yes - a MacOSX slave would be very useful, as I don't currently own  
> any
> Mac hardware.
>
> --
> Steve
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San  
> Francisco, CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the  
> Enterprise
> -Strategies to boost innovation and cut costs with open source  
> participation
> -Receive a $600 discount off the registration fee with the source  
> code: SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel




From dustin at zmanda.com  Thu Feb 19 21:06:33 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Thu, 19 Feb 2009 16:06:33 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <5E44F324-3AFE-440B-A009-63A9122ED5D0@mozilla.com>
References: <112350350473f2dba311.2089018456.stephen@starswan.com>
	<42338fbf0902190725w4566d6a2sfb0da3ea641a68eb@mail.gmail.com>
	<499DC130.70307@starswan.com>
	<5E44F324-3AFE-440B-A009-63A9122ED5D0@mozilla.com>
Message-ID: <42338fbf0902191306g3fc9229dp4593a89b3d0417b0@mail.gmail.com>

Great!

We also need a good supply of regular ol' Linux boxes .. who has some
spare cycles?

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From thomas at cranksoftware.com  Thu Feb 19 23:54:34 2009
From: thomas at cranksoftware.com (Thomas Fletcher)
Date: Thu, 19 Feb 2009 18:54:34 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <42338fbf0902191306g3fc9229dp4593a89b3d0417b0@mail.gmail.com>
References: <112350350473f2dba311.2089018456.stephen@starswan.com>	<42338fbf0902190725w4566d6a2sfb0da3ea641a68eb@mail.gmail.com>	<499DC130.70307@starswan.com>	<5E44F324-3AFE-440B-A009-63A9122ED5D0@mozilla.com>
	<42338fbf0902191306g3fc9229dp4593a89b3d0417b0@mail.gmail.com>
Message-ID: <499DF13A.6070802@cranksoftware.com>

Dustin J. Mitchell wrote:
> Great!
> 
> We also need a good supply of regular ol' Linux boxes .. who has some
> spare cycles?

We can probably help out with this in some capacity.

Thomas
-- 
Thomas Fletcher
Crank Software Inc.
www.cranksoftware.com



From bhearsum at mozilla.com  Thu Feb 19 23:56:49 2009
From: bhearsum at mozilla.com (Ben Hearsum)
Date: Thu, 19 Feb 2009 18:56:49 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <42338fbf0902181441h61a2b6b6s37c57ddab066d44d@mail.gmail.com>
References: <49930F38.2040400@starswan.com>
	<42338fbf0902111001w2195bfc7o6c8846d65f65cb5a@mail.gmail.com>
	<77fcfd0a0902111011i61f91402h6ef589d89f113b0e@mail.gmail.com>
	<42338fbf0902181441h61a2b6b6s37c57ddab066d44d@mail.gmail.com>
Message-ID: <0A8381C5-882B-4D35-ACF8-8EBEABCA04C9@mozilla.com>

Oh, I also meant to say:
If we're lacking a place to run the master I'm happy to do so on my  
VPS. That goes for any other hosting things too: e-mail/web/whatever

On 18-Feb-09, at 5:41 PM, Dustin J. Mitchell wrote:

> What's the status of this proposal?  I'd really like to get this set
> up, if possible in time to find and fix some bugs before the 0.7.10
> release!
>
> Dustin
>
> -- 
> Storage Software Engineer
> http://www.zmanda.com
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San  
> Francisco, CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the  
> Enterprise
> -Strategies to boost innovation and cut costs with open source  
> participation
> -Receive a $600 discount off the registration fee with the source  
> code: SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel




From dustin at zmanda.com  Fri Feb 20 01:46:49 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Thu, 19 Feb 2009 20:46:49 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <0A8381C5-882B-4D35-ACF8-8EBEABCA04C9@mozilla.com>
References: <49930F38.2040400@starswan.com>
	<42338fbf0902111001w2195bfc7o6c8846d65f65cb5a@mail.gmail.com>
	<77fcfd0a0902111011i61f91402h6ef589d89f113b0e@mail.gmail.com>
	<42338fbf0902181441h61a2b6b6s37c57ddab066d44d@mail.gmail.com>
	<0A8381C5-882B-4D35-ACF8-8EBEABCA04C9@mozilla.com>
Message-ID: <42338fbf0902191746w3412bd04me8a916c189816832@mail.gmail.com>

On Thu, Feb 19, 2009 at 6:56 PM, Ben Hearsum <bhearsum at mozilla.com> wrote:
> If we're lacking a place to run the master I'm happy to do so on my VPS.
> That goes for any other hosting things too: e-mail/web/whatever

Thanks for the offer!  At the moment, we're happy with sourceforge's
hosting for the mailing list (and using gmane for archives).  As for
web hosting, Brian will continue to host the Trac instance, and I'm
going to host generated HTML (documentation) on github.

Thanks, everyone -- it's good to know this is a team effort!

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From deejay1 at srem.org  Fri Feb 20 07:31:49 2009
From: deejay1 at srem.org (=?UTF-8?B?xYF1a2FzeiBKZXJuYcWb?=)
Date: Fri, 20 Feb 2009 08:31:49 +0100
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <42338fbf0902191306g3fc9229dp4593a89b3d0417b0@mail.gmail.com>
References: <112350350473f2dba311.2089018456.stephen@starswan.com>
	<42338fbf0902190725w4566d6a2sfb0da3ea641a68eb@mail.gmail.com>
	<499DC130.70307@starswan.com>
	<5E44F324-3AFE-440B-A009-63A9122ED5D0@mozilla.com>
	<42338fbf0902191306g3fc9229dp4593a89b3d0417b0@mail.gmail.com>
Message-ID: <77fcfd0a0902192331sfc1a380t2a52de8f8e411d62@mail.gmail.com>

On Thu, Feb 19, 2009 at 10:06 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
> Great!
>
> We also need a good supply of regular ol' Linux boxes .. who has some
> spare cycles?

Like I said I have one lying around ;)

-- 
?ukasz



From Gregory.Manning at mitrefinch.co.uk  Fri Feb 20 09:35:56 2009
From: Gregory.Manning at mitrefinch.co.uk (Gregory Manning)
Date: Fri, 20 Feb 2009 09:35:56 +0000
Subject: [Buildbot-devel] The Meta Buildbot
Message-ID: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>

I have a MacOSX (10.5) desktop machine which is on more than off, I'm happy to offer that as a buildslave.
My linux server is probably available too.

Greg Manning
Development Analyst
Mitrefinch Ltd.
 
 

This message has been scanned for malware by Websense. www.websense.com



From mahatmamanic at gmail.com  Sat Feb 21 02:15:31 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Fri, 20 Feb 2009 18:15:31 -0800
Subject: [Buildbot-devel] svn server certificates
In-Reply-To: <c02dbb2a0902201751s32da0046r36513aa818315041@mail.gmail.com>
References: <c02dbb2a0902201751s32da0046r36513aa818315041@mail.gmail.com>
Message-ID: <c02dbb2a0902201815p737d2480g2ef21bd25c2b6979@mail.gmail.com>

Please ignore.  I was a bit frantic as it's Friday night, a build had been
requested for QA to test over the weekend, and I was frustrated.  The
problem was of course that it doesn't matter what has been accepted on the
buildmaster, if the certificate has not been accepted on the buildslave.
*embarrassed*

Sorry for the extra traffic.

Ian


On Fri, Feb 20, 2009 at 5:51 PM, Ian Peters-Campbell <mahatmamanic at gmail.com
> wrote:

> Hey all,
>
> I am having some trouble with SVn rejecting server certificates.  I've
> manually accepted the certificate on the command line for the user the
> buildmaster is running under, and I am able to use SVN from the command line
> with no problem.  In fact, just two days ago buildbot was running fine and
> chuckling along with svn checkouts, but now all the builds are failing due
> to rejected certificates.  Has any oine experienced this issue, or have any
> idea how to deal with it?
>
> Ian
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090220/f17baf41/attachment.html>

From mahatmamanic at gmail.com  Sat Feb 21 01:51:38 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Fri, 20 Feb 2009 17:51:38 -0800
Subject: [Buildbot-devel] svn server certificates
Message-ID: <c02dbb2a0902201751s32da0046r36513aa818315041@mail.gmail.com>

Hey all,

I am having some trouble with SVn rejecting server certificates.  I've
manually accepted the certificate on the command line for the user the
buildmaster is running under, and I am able to use SVN from the command line
with no problem.  In fact, just two days ago buildbot was running fine and
chuckling along with svn checkouts, but now all the builds are failing due
to rejected certificates.  Has any oine experienced this issue, or have any
idea how to deal with it?

Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090220/88e6753e/attachment.html>

From stephen at starswan.com  Sun Feb 22 19:23:56 2009
From: stephen at starswan.com (Stephen Dicks)
Date: Sun, 22 Feb 2009 19:23:56 +0000
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>
References: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>
Message-ID: <49A1A64C.4070401@starswan.com>

I now have a basic meta-buildbot running off dustin's main git
repository at:

http://buildbot.starswan.com/

it would be good to add some other slaves - could people mail me
off-list with their configurations and slave names/passwords?

However I can't seem to run the tests - it seems to want to write to
site-packages in the installed python directory - should it really be
doing that?


--
Steve Dicks



From l10n.moz at googlemail.com  Sun Feb 22 19:59:45 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Sun, 22 Feb 2009 20:59:45 +0100
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <49A1A64C.4070401@starswan.com>
References: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>
	<49A1A64C.4070401@starswan.com>
Message-ID: <9de1bdb60902221159x5e131bf4yd22f97cc1befa526@mail.gmail.com>

I'm seeing similar warnings when I run my own twisted plugin. But that's
really just a warning, the daemon does indeed run.

Not that I know why it's trying to run a plugin. Maybe the reporter stuff?

Axel

2009/2/22 Stephen Dicks <stephen at starswan.com>

> I now have a basic meta-buildbot running off dustin's main git
> repository at:
>
> http://buildbot.starswan.com/
>
> it would be good to add some other slaves - could people mail me
> off-list with their configurations and slave names/passwords?
>
> However I can't seem to run the tests - it seems to want to write to
> site-packages in the installed python directory - should it really be
> doing that?
>
>
> --
> Steve Dicks
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
> CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the
> Enterprise
> -Strategies to boost innovation and cut costs with open source
> participation
> -Receive a $600 discount off the registration fee with the source code:
> SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090222/a82863d5/attachment.html>

From stephen at starswan.com  Sun Feb 22 20:08:30 2009
From: stephen at starswan.com (Stephen Dicks)
Date: Sun, 22 Feb 2009 20:08:30 +0000
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <9de1bdb60902221159x5e131bf4yd22f97cc1befa526@mail.gmail.com>
References: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>	
	<49A1A64C.4070401@starswan.com>
	<9de1bdb60902221159x5e131bf4yd22f97cc1befa526@mail.gmail.com>
Message-ID: <49A1B0BE.1010705@starswan.com>

Axel Hecht wrote:
> I'm seeing similar warnings when I run my own twisted plugin. But
> that's really just a warning, the daemon does indeed run.
>
> Not that I know why it's trying to run a plugin. Maybe the reporter stuff?
>
Well here it seems to fail everything - its a permissions violation
writing to the directory.

--
Steve



From l10n.moz at googlemail.com  Sun Feb 22 20:35:30 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Sun, 22 Feb 2009 21:35:30 +0100
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <49A1B0BE.1010705@starswan.com>
References: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>
	<49A1A64C.4070401@starswan.com>
	<9de1bdb60902221159x5e131bf4yd22f97cc1befa526@mail.gmail.com>
	<49A1B0BE.1010705@starswan.com>
Message-ID: <9de1bdb60902221235o37024336h708a8f5e493472c3@mail.gmail.com>

2009/2/22 Stephen Dicks <stephen at starswan.com>

> Axel Hecht wrote:
> > I'm seeing similar warnings when I run my own twisted plugin. But
> > that's really just a warning, the daemon does indeed run.
> >
> > Not that I know why it's trying to run a plugin. Maybe the reporter
> stuff?
> >
> Well here it seems to fail everything - its a permissions violation
> writing to the directory.
>

Yes, and the thing that's writing is twisted caching a plugin that it
doesn't know about by install.

Does trial pass on the commandline?

Did you specify tests in the step?

Axel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090222/c78cef67/attachment.html>

From dustin at zmanda.com  Mon Feb 23 00:09:57 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Sun, 22 Feb 2009 19:09:57 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <49A1A64C.4070401@starswan.com>
References: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>
	<49A1A64C.4070401@starswan.com>
Message-ID: <42338fbf0902221609n5d0fe6e3s3e51985e0acc582c@mail.gmail.com>

On Sun, Feb 22, 2009 at 2:23 PM, Stephen Dicks <stephen at starswan.com> wrote:
> I now have a basic meta-buildbot running off dustin's main git
> repository at:
>
> http://buildbot.starswan.com/

Awesome!  Thanks!

> it would be good to add some other slaves - could people mail me
> off-list with their configurations and slave names/passwords?

Will do.

> However I can't seem to run the tests - it seems to want to write to
> site-packages in the installed python directory - should it really be
> doing that?

Hmm -- the exception is:
Traceback (most recent call last):
  File "/usr/lib/python2.4/site-packages/twisted/python/usage.py",
line 241, in parseOptions
    self.postOptions()
  File "/usr/lib/python2.4/site-packages/twisted/scripts/trial.py",
line 293, in postOptions
    self['reporter'] = self._loadReporterByName(self['reporter'])
  File "/usr/lib/python2.4/site-packages/twisted/scripts/trial.py",
line 279, in _loadReporterByName
    for p in plugin.getPlugins(itrial.IReporter):
  File "/usr/lib/python2.4/site-packages/twisted/plugin.py", line 200,
in getPlugins
    allDropins = getCache(package)
--- <exception caught here> ---
  File "/usr/lib/python2.4/site-packages/twisted/plugin.py", line 179,
in getCache
    dropinPath.setContent(pickle.dumps(dropinDotCache))
  File "/usr/lib/python2.4/site-packages/twisted/python/filepath.py",
line 574, in setContent
    sib.open('w').write(content)
  File "/usr/lib/python2.4/site-packages/twisted/python/filepath.py",
line 384, in open
    return open(self.path, mode+'b')
exceptions.IOError: [Errno 13] Permission denied:
'/usr/lib/python2.4/site-packages/twisted/plugins/dropin.cache.new'

which looks like a twisted thing to me, rather than a test failure.
Buildbot and Twisted hacker Jean-Paul (exarkun) pointed out the
solution here:
  http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=477103
Basically -- run 'twistd --help' as root, and this problem will be solved.

It's the failures at the end that I worry about, but I think I've
fixed those and pushed the changes.  We'll see :)

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From mahatmamanic at gmail.com  Mon Feb 23 23:01:14 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Mon, 23 Feb 2009 15:01:14 -0800
Subject: [Buildbot-devel] Buildslave status mismatch between web/IRC
Message-ID: <c02dbb2a0902231501g6a6bdad4w268bf131bc56e77c@mail.gmail.com>

I've been running into an odd thing lately, where it seems that I have my
build slaves configured correctly, and the waterfall page shows the
buildslaves are online and attached to the proper builders, but my IRC bots
are reporting my builders as being offline.  When I attempt to order a build
they explain that it is because I don't have any buildslaves up.

I'm certain that this is some configuration tweak I have wrong, and I know
I'll be kicking myself when I know what the problem is.  I'm hoping someone
has some insight as to what I am doing wrong.

Thanks,
Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090223/526f2b46/attachment.html>

From seb at highlab.com  Tue Feb 24 07:09:54 2009
From: seb at highlab.com (Sebastian Kuzminsky)
Date: Tue, 24 Feb 2009 00:09:54 -0700
Subject: [Buildbot-devel] Buildslave status mismatch between web/IRC
In-Reply-To: <c02dbb2a0902231501g6a6bdad4w268bf131bc56e77c@mail.gmail.com>
References: <c02dbb2a0902231501g6a6bdad4w268bf131bc56e77c@mail.gmail.com>
Message-ID: <49A39D42.6010207@highlab.com>

Ian Peters-Campbell wrote:
> I've been running into an odd thing lately, where it seems that I have my
> build slaves configured correctly, and the waterfall page shows the
> buildslaves are online and attached to the proper builders, but my IRC bots
> are reporting my builders as being offline.  When I attempt to order a build
> they explain that it is because I don't have any buildslaves up.
> 
> I'm certain that this is some configuration tweak I have wrong, and I know
> I'll be kicking myself when I know what the problem is.  I'm hoping someone
> has some insight as to what I am doing wrong.

Do you mean that you're talking on IRC to the buildmaster's IRC bot, and 
it doesnt know about your buildslaves?

If that's the case, are you maybe talking to the wrong buildmaster? 
There is a bunch of them on IRC.  Does your buildbot have an 
undistinguished IRC nick like "buildbot"?


-- 
Sebastian Kuzminsky
maybe you should be spending more time thinking about your monkey
<http://www.youtube.com/watch?v=ZT6qHbEL68M>



From mahatmamanic at gmail.com  Tue Feb 24 14:20:18 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Tue, 24 Feb 2009 06:20:18 -0800
Subject: [Buildbot-devel] Buildslave status mismatch between web/IRC
In-Reply-To: <49A39D42.6010207@highlab.com>
References: <c02dbb2a0902231501g6a6bdad4w268bf131bc56e77c@mail.gmail.com>
	<49A39D42.6010207@highlab.com>
Message-ID: <c02dbb2a0902240620n5cbab9a2rc89be121f3d1aac2@mail.gmail.com>

Well, I have a single buildmaster running on a private network, with a
private, dedicated IRC server.  The IRC bots have names of <Project>Bot and
live in #<Project> chat rooms.  Each bot has a category of <Project> and
each builder within the project (generally Integration, QA, and Release)
also has a category of <Project>

When queried with something like ProjectBot: list builders the bot reponds
with something along the lines of:

Project-debug<offline> Project-release<offline> Project-integration<offline>

And if I attempt to force a build it throws an interfaces.NoSlave exception
from within IRCContact.commandFORCE

Meanwhile, on the waterfall page the build slaves are reporting in, map to
builders properly, and are listing the admin/hardware info I would expect.

A possible solution came to me when I woke up this morning that i will
explore when i get into work.  I am wondering if the builder is reporting
offline if I have not manually created its builddir on at least one of its
online buildslaves,   For some reason  Ihave a niggling suspicion this
morning that if I manually create those build directories the builders wil
start recognizing that they have valid slaves and will come online.  Once i
can check that out at work I will report in with my results :)

Ian


On Mon, Feb 23, 2009 at 11:09 PM, Sebastian Kuzminsky <seb at highlab.com>wrote:

> Ian Peters-Campbell wrote:
>
>> I've been running into an odd thing lately, where it seems that I have my
>> build slaves configured correctly, and the waterfall page shows the
>> buildslaves are online and attached to the proper builders, but my IRC
>> bots
>> are reporting my builders as being offline.  When I attempt to order a
>> build
>> they explain that it is because I don't have any buildslaves up.
>>
>> I'm certain that this is some configuration tweak I have wrong, and I know
>> I'll be kicking myself when I know what the problem is.  I'm hoping
>> someone
>> has some insight as to what I am doing wrong.
>>
>
> Do you mean that you're talking on IRC to the buildmaster's IRC bot, and it
> doesnt know about your buildslaves?
>
> If that's the case, are you maybe talking to the wrong buildmaster? There
> is a bunch of them on IRC.  Does your buildbot have an undistinguished IRC
> nick like "buildbot"?
>
>
> --
> Sebastian Kuzminsky
> maybe you should be spending more time thinking about your monkey
> <http://www.youtube.com/watch?v=ZT6qHbEL68M>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090224/47f2872a/attachment.html>

From mahatmamanic at gmail.com  Tue Feb 24 17:50:54 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Tue, 24 Feb 2009 09:50:54 -0800
Subject: [Buildbot-devel] Buildslave status mismatch between web/IRC
In-Reply-To: <c02dbb2a0902240620n5cbab9a2rc89be121f3d1aac2@mail.gmail.com>
References: <c02dbb2a0902231501g6a6bdad4w268bf131bc56e77c@mail.gmail.com>
	<49A39D42.6010207@highlab.com>
	<c02dbb2a0902240620n5cbab9a2rc89be121f3d1aac2@mail.gmail.com>
Message-ID: <c02dbb2a0902240950n608897faj3ec645c7ae1c763e@mail.gmail.com>

Victory :)

Ok so I just verified that the way to fix the problem is to manually create
the builder's builddir path on the buildslave and then restart both the
buildmaster and the buildslave (probably just the buildslave, but I
restarted both.)

It might be handy if buildbot could either
A) automagically create the relative builddir for a builder if it does not
exist or
B) Do some more granular checking as to the buildslave problem rather than
simply reporting that the buildslave is offline

I'm no Twisted expert so I am not sure how much help I would be in
implementing something like that, but I'd be happy to try and cut my teeth
on it if it's possible and someone could point me to the right portion of
the code to look at. :)

Ian


On Tue, Feb 24, 2009 at 6:20 AM, Ian Peters-Campbell <mahatmamanic at gmail.com
> wrote:

> Well, I have a single buildmaster running on a private network, with a
> private, dedicated IRC server.  The IRC bots have names of <Project>Bot and
> live in #<Project> chat rooms.  Each bot has a category of <Project> and
> each builder within the project (generally Integration, QA, and Release)
> also has a category of <Project>
>
> When queried with something like ProjectBot: list builders the bot reponds
> with something along the lines of:
>
> Project-debug<offline> Project-release<offline>
> Project-integration<offline>
>
> And if I attempt to force a build it throws an interfaces.NoSlave exception
> from within IRCContact.commandFORCE
>
> Meanwhile, on the waterfall page the build slaves are reporting in, map to
> builders properly, and are listing the admin/hardware info I would expect.
>
> A possible solution came to me when I woke up this morning that i will
> explore when i get into work.  I am wondering if the builder is reporting
> offline if I have not manually created its builddir on at least one of its
> online buildslaves,   For some reason  Ihave a niggling suspicion this
> morning that if I manually create those build directories the builders wil
> start recognizing that they have valid slaves and will come online.  Once i
> can check that out at work I will report in with my results :)
>
> Ian
>
>
>
> On Mon, Feb 23, 2009 at 11:09 PM, Sebastian Kuzminsky <seb at highlab.com>wrote:
>
>> Ian Peters-Campbell wrote:
>>
>>> I've been running into an odd thing lately, where it seems that I have my
>>> build slaves configured correctly, and the waterfall page shows the
>>> buildslaves are online and attached to the proper builders, but my IRC
>>> bots
>>> are reporting my builders as being offline.  When I attempt to order a
>>> build
>>> they explain that it is because I don't have any buildslaves up.
>>>
>>> I'm certain that this is some configuration tweak I have wrong, and I
>>> know
>>> I'll be kicking myself when I know what the problem is.  I'm hoping
>>> someone
>>> has some insight as to what I am doing wrong.
>>>
>>
>> Do you mean that you're talking on IRC to the buildmaster's IRC bot, and
>> it doesnt know about your buildslaves?
>>
>> If that's the case, are you maybe talking to the wrong buildmaster? There
>> is a bunch of them on IRC.  Does your buildbot have an undistinguished IRC
>> nick like "buildbot"?
>>
>>
>> --
>> Sebastian Kuzminsky
>> maybe you should be spending more time thinking about your monkey
>> <http://www.youtube.com/watch?v=ZT6qHbEL68M>
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090224/1b33bea5/attachment.html>

From mahatmamanic at gmail.com  Tue Feb 24 18:10:36 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Tue, 24 Feb 2009 10:10:36 -0800
Subject: [Buildbot-devel] Build/BuildStep getProperties w/ defaults?
Message-ID: <c02dbb2a0902241010w54ebe47sa6fd1c5b45c246bc@mail.gmail.com>

Has anyone written a getProperties for builds and buildsteps that accepts
default values?  If not I will add some in manually unless there's an
establihed reason not to do so?  It looks like Properties.getProperty() is
already capable of accepting default values, so it should be pretty minor?

Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090224/ce3005cb/attachment.html>

From dustin at zmanda.com  Tue Feb 24 18:22:03 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Tue, 24 Feb 2009 13:22:03 -0500
Subject: [Buildbot-devel] Build/BuildStep getProperties w/ defaults?
In-Reply-To: <c02dbb2a0902241010w54ebe47sa6fd1c5b45c246bc@mail.gmail.com>
References: <c02dbb2a0902241010w54ebe47sa6fd1c5b45c246bc@mail.gmail.com>
Message-ID: <42338fbf0902241022q3b1aab46w52d4f6ec7f148d03@mail.gmail.com>

On Tue, Feb 24, 2009 at 1:10 PM, Ian Peters-Campbell
<mahatmamanic at gmail.com> wrote:
> Has anyone written a getProperties for builds and buildsteps that accepts
> default values?? If not I will add some in manually unless there's an
> establihed reason not to do so?? It looks like Properties.getProperty() is
> already capable of accepting default values, so it should be pretty minor?

Sounds fine to me.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From claude.vittoria at orange-ftgroup.com  Wed Feb 25 07:46:27 2009
From: claude.vittoria at orange-ftgroup.com (VITTORIA Claude BU IT&L@bs)
Date: Wed, 25 Feb 2009 08:46:27 +0100
Subject: [Buildbot-devel] My doxygen step
Message-ID: <1235547987.3415.7.camel@MemphisBell.silicomp.fr>

Hi,

I wrote some time ago my "step" to transfere generated files in a tar
archive and to untar on the master system.
I wrote that to solve a problem of bandwith (VPN between site and ssh
forbidden)and short window time to transfer (not sure that 393 solves my
problem). The buildmaster server also is my web server (same storage but
different systems in reality).

The following code has no pretension, I wrote to solve a problem and I
share an experience. The code below has been written for 0.7.8. All
ideas to improve are wellcome. The main issue, I suppose, is that master
is "blocked" during untar step, not fair if an heavy load, how to
"thread" it ?

Claude
===========================================================================
Master step only
===========================================================================
class untar(BuildStep):

    def __init__ (self, tarpath='*', tarmask='',
tardescription='Untar',**buildstep_kwargs):
        self.tarpath=tarpath
        self.tarmask=tarmask
        self.tardescription=tardescription
        BuildStep.__init__(self, **buildstep_kwargs)
        self.addFactoryArguments(tarpath=tarpath,
                                 tarmask=tarmask,
                                 tardescription=tardescription)
        
    def start(self):
        source = os.path.expanduser(self.tarpath)
        self.step_status.setColor('yellow')
        self.step_status.setText(['Inflating'])
        status = os.system('tar xvf "%s"'%(source))
        if self.tarmask != '':
            sourcech = os.path.expanduser(self.tarmask)
            status = os.system('chmod o+rx "%s"'%(sourcech))
            status = os.system('chmod -R o+r "%s"'%(sourcech))

        if status == 0:
            self.step_status.setColor('green')
            self.step_status.setText([self.tardescription,
os.path.basename(source)])
            return BuildStep.finished(self, SUCCESS)
        else:
            self.step_status.setColor('red')
            self.step_status.setText(["Untar",
os.path.basename(source)])
            return BuildStep.finished(self, FAILURE)
===================================================================================
factory conf
===================================================================================
f.addStep(ShellCommand(command=["make", "doc"])),
f.addStep(ShellCommand(command=["tar", "cvf", "doxygen.tar",
"doc/html"], haltOnFailure = 1)),
f.addStep(FileUpload(slavesrc="doxygen.tar",
                          masterdest="~buildmaster/doc/doxygen.tar")),
f.addStep(untar(tarpath="~buildmaster/doc/doxygen.tar",
tarmask="~buildmaster/doc/html", tardescription="<a href=
\"http://buildbot/AIS007C1/doxygen\">Doxygen</a>"))




From dustin at zmanda.com  Wed Feb 25 15:54:32 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 25 Feb 2009 10:54:32 -0500
Subject: [Buildbot-devel] My doxygen step
In-Reply-To: <1235547987.3415.7.camel@MemphisBell.silicomp.fr>
References: <1235547987.3415.7.camel@MemphisBell.silicomp.fr>
Message-ID: <42338fbf0902250754v1ce722bp4ff8c246a941dab5@mail.gmail.com>

I think MasterShellCommand is what you need:
  http://github.com/djmitche/buildbot/blob/master/buildbot/steps/master.py

-- 
Storage Software Engineer
http://www.zmanda.com



From dbailey at hp.com  Wed Feb 25 16:48:37 2009
From: dbailey at hp.com (Bailey, Darragh)
Date: Wed, 25 Feb 2009 16:48:37 +0000
Subject: [Buildbot-devel] Renaming builds and retaining history
Message-ID: <FF452E17060D614D89BF7FD3A07A88333202E0B9D4@GVW1115EXC.americas.hpqcorp.net>


Is it possible to rename a build and retain it's history?

I've got a couple of builds that tbh, I should have named them entirely differently, but if I rename them in the config file I'll lose their entire build history.


--
Regards,
Darragh Bailey

Systems Software Engineer
Hewlett Packard Galway Ltd.

Postal Address:??? Hewlett Packard Galway Limited, Ballybrit Business Park, Galway
Registered Office: Hewlett Packard Galway Limited, 63-74 Sir John Rogerson's Quay Dublin 2
Registered Number: 361933 


From dustin at zmanda.com  Wed Feb 25 16:59:00 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 25 Feb 2009 11:59:00 -0500
Subject: [Buildbot-devel] Renaming builds and retaining history
In-Reply-To: <FF452E17060D614D89BF7FD3A07A88333202E0B9D4@GVW1115EXC.americas.hpqcorp.net>
References: <FF452E17060D614D89BF7FD3A07A88333202E0B9D4@GVW1115EXC.americas.hpqcorp.net>
Message-ID: <42338fbf0902250859l436fbc52i3a1adcf140ed0a@mail.gmail.com>

On Wed, Feb 25, 2009 at 11:48 AM, Bailey, Darragh <dbailey at hp.com> wrote:
> Is it possible to rename a build and retain it's history?
>
> I've got a couple of builds that tbh, I should have named them entirely differently, but if I rename them in the config file I'll lose their entire build history.

If you also rename the pickles in the master directory, I *think*
things will work out.  No promises, though!

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From l10n.moz at googlemail.com  Wed Feb 25 17:08:15 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Wed, 25 Feb 2009 18:08:15 +0100
Subject: [Buildbot-devel] step_status.setColor gone in git master,
	worth the breakage?
Message-ID: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>

Hi,

I just noticed that
http://github.com/djmitche/buildbot/commit/d1190bd358845fe38204e89d8399f95a83309334#diff-7removed
step_status.setColor. I have one step, and the mozilla guys have a
flock of other steps, that don't take that change well.

I wonder if we should keep that method as a no-op, or possibly log.msg()
about it not doing anything anymore.

Right now, it seems that custom steps need to make a call on whether they
run up to 0.7.9 or on 0.7.10.

Axel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090225/2dfe96a3/attachment.html>

From neil.hemingway at greyhavens.org.uk  Wed Feb 25 17:20:13 2009
From: neil.hemingway at greyhavens.org.uk (Neil Hemingway)
Date: Wed, 25 Feb 2009 17:20:13 +0000
Subject: [Buildbot-devel] step_status.setColor gone in git master,
	worth 	the breakage?
In-Reply-To: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>
References: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>
Message-ID: <47be741a0902250920y557bcd05j704d6db0df379e79@mail.gmail.com>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

That's not good.

I removed all the colour stuff as everything that cared set the status
anyway, and it was better to have the colour set by CSS classes (which
were already being set based on the status)

I have no problem with the method being stubbed out, but I wouldn't
want it to be reverted, as this gives the misleading impression that
it's useful.

I think it would probably be best to have it log.msg() that it's been
deprecated.

Is it just BuildStepStatus.setColor that's a problem or elsewhere (I
removed a bunch)?

Neil


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkmlfcQACgkQZjaLPuxR95Q8qgCeN1FKOIcsNoXZWDq56Pq8W9DB
kMcAoJoT9AFq5fuCBHIU18BryBM8RHXq
=WX+q
-----END PGP SIGNATURE-----
2009/2/25 Axel Hecht <l10n.moz at googlemail.com>

> Hi,
>
> I just noticed that
> http://github.com/djmitche/buildbot/commit/d1190bd358845fe38204e89d8399f95a83309334#diff-7removed step_status.setColor. I have one step, and the mozilla guys have a
> flock of other steps, that don't take that change well.
>
> I wonder if we should keep that method as a no-op, or possibly log.msg()
> about it not doing anything anymore.
>
> Right now, it seems that custom steps need to make a call on whether they
> run up to 0.7.9 or on 0.7.10.
>
> Axel
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
> CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the
> Enterprise
> -Strategies to boost innovation and cut costs with open source
> participation
> -Receive a $600 discount off the registration fee with the source code:
> SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090225/ac42370b/attachment.html>

From dustin at zmanda.com  Wed Feb 25 17:21:21 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 25 Feb 2009 12:21:21 -0500
Subject: [Buildbot-devel] step_status.setColor gone in git master,
	worth 	the breakage?
In-Reply-To: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>
References: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>
Message-ID: <42338fbf0902250921g7167a3dfg30ec4051354d0a5e@mail.gmail.com>

On Wed, Feb 25, 2009 at 12:08 PM, Axel Hecht <l10n.moz at googlemail.com> wrote:
> I just noticed that
> http://github.com/djmitche/buildbot/commit/d1190bd358845fe38204e89d8399f95a83309334#diff-7
> removed step_status.setColor. I have one step, and the mozilla guys have a
> flock of other steps, that don't take that change well.
>
> I wonder if we should keep that method as a no-op, or possibly log.msg()
> about it not doing anything anymore.
>
> Right now, it seems that custom steps need to make a call on whether they
> run up to 0.7.9 or on 0.7.10.

Hmm, is it *that* bad?  This is the problem with vague
"read-the-source-and-do-what-seems-best" APIs :(

It's not clear to me that a no-op (with attendant disappearance of
color) is better than a traceback.  But I'm open to be convinced.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Wed Feb 25 17:36:59 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Wed, 25 Feb 2009 12:36:59 -0500
Subject: [Buildbot-devel] step_status.setColor gone in git master,
	worth 	the breakage?
In-Reply-To: <42338fbf0902250921g7167a3dfg30ec4051354d0a5e@mail.gmail.com>
References: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>
	<42338fbf0902250921g7167a3dfg30ec4051354d0a5e@mail.gmail.com>
Message-ID: <42338fbf0902250936id1b867icac8058eac1d6968@mail.gmail.com>

On Wed, Feb 25, 2009 at 12:21 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
> It's not clear to me that a no-op (with attendant disappearance of
> color) is better than a traceback. ?But I'm open to be convinced.

Just created:
  http://buildbot.net/trac/ticket/435

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From l10n.moz at googlemail.com  Wed Feb 25 18:08:47 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Wed, 25 Feb 2009 19:08:47 +0100
Subject: [Buildbot-devel] step_status.setColor gone in git master,
	worth 	the breakage?
In-Reply-To: <42338fbf0902250936id1b867icac8058eac1d6968@mail.gmail.com>
References: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>
	<42338fbf0902250921g7167a3dfg30ec4051354d0a5e@mail.gmail.com>
	<42338fbf0902250936id1b867icac8058eac1d6968@mail.gmail.com>
Message-ID: <9de1bdb60902251008m2e17a966kb4a99325a0342d9e@mail.gmail.com>

Filed https://bugzilla.mozilla.org/show_bug.cgi?id=480145 on the Mozilla
steps part.

Nevertheless, I think that adding a noop method on BuildStepStatus is an
easy win in terms of letting people migrate to .10.

/me notes that he can't type 10 without a 10n, that's what I get from being
l10n.

Axel

2009/2/25 Dustin J. Mitchell <dustin at zmanda.com>

> On Wed, Feb 25, 2009 at 12:21 PM, Dustin J. Mitchell <dustin at zmanda.com>
> wrote:
> > It's not clear to me that a no-op (with attendant disappearance of
> > color) is better than a traceback.  But I'm open to be convinced.
>
> Just created:
>  http://buildbot.net/trac/ticket/435
>
> Dustin
>
> --
> Storage Software Engineer
> http://www.zmanda.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090225/fdea4f6a/attachment.html>

From neil.hemingway at greyhavens.org.uk  Wed Feb 25 18:26:45 2009
From: neil.hemingway at greyhavens.org.uk (Neil Hemingway)
Date: Wed, 25 Feb 2009 18:26:45 +0000
Subject: [Buildbot-devel] step_status.setColor gone in git master,
	worth 	the breakage?
In-Reply-To: <42338fbf0902250921g7167a3dfg30ec4051354d0a5e@mail.gmail.com>
References: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>
	<42338fbf0902250921g7167a3dfg30ec4051354d0a5e@mail.gmail.com>
Message-ID: <47be741a0902251026p19ab4410k87d9225f78afd62c@mail.gmail.com>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

2009/2/25 Dustin J. Mitchell <dustin at zmanda.com>

It's not clear to me that a no-op (with attendant disappearance of
color) is better than a traceback.  But I'm open to be convinced.

The core buildbot code doesn't use the color attribute (hence my
removing it).  All the HTML stuff used CSS classes based on the
status, the only thing that still has color embedded is the guistatus
gtk app.

So the only reason that color should disappear for anyone if we stub
out setColor(), is if their custom code doesn't also call setStatus()

Neil

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkmljU4ACgkQZjaLPuxR95RoagCfRlRL+TjjFi/e5Ho8ytwix8Ny
U1MAn3NlBpz81htUm+6DQ0jhzWs//v0X
=AkHp
-----END PGP SIGNATURE-----
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090225/4d125898/attachment.html>

From l10n.moz at googlemail.com  Wed Feb 25 19:59:37 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Wed, 25 Feb 2009 20:59:37 +0100
Subject: [Buildbot-devel] step_status.setColor gone in git master,
	worth 	the breakage?
In-Reply-To: <47be741a0902251026p19ab4410k87d9225f78afd62c@mail.gmail.com>
References: <9de1bdb60902250908q7c9783c7va309017c26224b66@mail.gmail.com>
	<42338fbf0902250921g7167a3dfg30ec4051354d0a5e@mail.gmail.com>
	<47be741a0902251026p19ab4410k87d9225f78afd62c@mail.gmail.com>
Message-ID: <9de1bdb60902251159i6cf7c41j4bedb26f2907b4a7@mail.gmail.com>

2009/2/25 Neil Hemingway <neil.hemingway at greyhavens.org.uk>

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> 2009/2/25 Dustin J. Mitchell <dustin at zmanda.com>
>
> It's not clear to me that a no-op (with attendant disappearance of
> color) is better than a traceback.  But I'm open to be convinced.
>
> The core buildbot code doesn't use the color attribute (hence my
> removing it).  All the HTML stuff used CSS classes based on the
> status, the only thing that still has color embedded is the guistatus
> gtk app.
>
> So the only reason that color should disappear for anyone if we stub
> out setColor(), is if their custom code doesn't also call setStatus()
>

FYI, the ticket is closed and we have a fill-in method now that writes to
the log.

I suggest that you try to remove your use of setColor() in your custom steps
nevertheless, I expect that wallpaper to be torn down in a future release.

Thanks

Axel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090225/e4bdc945/attachment.html>

From drewdavenport.dev at gmail.com  Thu Feb 26 00:46:22 2009
From: drewdavenport.dev at gmail.com (Drew Davenport)
Date: Wed, 25 Feb 2009 16:46:22 -0800
Subject: [Buildbot-devel] Dynamically evaluated step status
Message-ID: <4f70e4db0902251646j4ef8fa47ve271be037ac365ba@mail.gmail.com>

Hello,

In one of my build processes, I would like to mark a step as passed/failed
based on the results of another action that is operating from outside
buildbot. In more concrete terms, the pass/fail status of a step is
determined by querying certain data in a database, and this data is updated
based on tests run on the results of this build. However, this information
is not immediately available when a build has completed, and I do not want
to block the builder waiting for the status to become available.

A person looking at the waterfall status could therefore see a completed
build's status change from <some colour> to <red/green> after refreshing the
page.

Is this possible? I have done some digging and it looks like the logs/status
are pretty much static once a build is complete. I am investigating how I
might bolt this on, but any guidance would be greatly appreciated.

Cheers,
Drew Davenport
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090225/0a7fee4c/attachment.html>

From dustin at zmanda.com  Thu Feb 26 14:45:37 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Thu, 26 Feb 2009 09:45:37 -0500
Subject: [Buildbot-devel] Dynamically evaluated step status
In-Reply-To: <4f70e4db0902251646j4ef8fa47ve271be037ac365ba@mail.gmail.com>
References: <4f70e4db0902251646j4ef8fa47ve271be037ac365ba@mail.gmail.com>
Message-ID: <42338fbf0902260645x28ac28a7pbe90d9e2c93ffa71@mail.gmail.com>

On Wed, Feb 25, 2009 at 7:46 PM, Drew Davenport
<drewdavenport.dev at gmail.com> wrote:
> Is this possible? I have done some digging and it looks like the logs/status
> are pretty much static once a build is complete. I am investigating how I
> might bolt this on, but any guidance would be greatly appreciated.

I don't think that changing the status after a step has completed is
practical, but certainly you can poll this "external process" in a
build step and reflects its outcome in the pass/fail status of the
build test.  Something as simple as
  ShellCommand(command="./poll-external-process.sh")
could do the trick.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From deejay1 at srem.org  Thu Feb 26 16:33:40 2009
From: deejay1 at srem.org (=?UTF-8?B?xYF1a2FzeiBKZXJuYcWb?=)
Date: Thu, 26 Feb 2009 17:33:40 +0100
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <42338fbf0902221609n5d0fe6e3s3e51985e0acc582c@mail.gmail.com>
References: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>
	<49A1A64C.4070401@starswan.com>
	<42338fbf0902221609n5d0fe6e3s3e51985e0acc582c@mail.gmail.com>
Message-ID: <77fcfd0a0902260833v2fd9e402w109fd1642014f467@mail.gmail.com>

Hi.

Does anyone have an idea what's going on with the meta buildbot? It
can't get the git tree...

-- 
?ukasz



From dustin at zmanda.com  Thu Feb 26 16:45:30 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Thu, 26 Feb 2009 11:45:30 -0500
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <77fcfd0a0902260833v2fd9e402w109fd1642014f467@mail.gmail.com>
References: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>
	<49A1A64C.4070401@starswan.com>
	<42338fbf0902221609n5d0fe6e3s3e51985e0acc582c@mail.gmail.com>
	<77fcfd0a0902260833v2fd9e402w109fd1642014f467@mail.gmail.com>
Message-ID: <42338fbf0902260845r7c671c42p44c1caa1f2dda782@mail.gmail.com>

On Thu, Feb 26, 2009 at 11:33 AM, ?ukasz Jerna? <deejay1 at srem.org> wrote:
> Does anyone have an idea what's going on with the meta buildbot? It
> can't get the git tree...

There's a bug with pulling via HTTP from github:
  http://support.github.com/discussions/repos/199-git-repo-missing-a-blob

In the interim, I think that we should be pulling from
  http://repo.or.cz/r/buildbot.git
but I haven't heard back from Stephen in a while.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Thu Feb 26 18:16:30 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Thu, 26 Feb 2009 13:16:30 -0500
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
Message-ID: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>

I've just finished releasing buildbot-0.7.10, available for download
in the usual places:

 http://pypi.python.org/pypi/buildbot/0.7.10
 https://sourceforge.net/project/showfiles.php?group_id=73177

File checksums are as follows.  This buildbot release is signed by my
GPG public key (7F0D15B1) (available from keyservers):

 buildbot-0.7.10.tar.gz
  md5: 3a51f6ba8a852382863d1940cef89a46

 buildbot-0.7.10.zip
  md5: 33946b82706e8b65a1f8dab2de7ab9bf

This release has *truly* been a community effort -- I did little more
than marshall patches from a number of contributors and developers.
Thank you to everyone who helped to make this release a reality, and
especially to those who pitched in over the last week or so to quash
the last few bugs.

----

This release is mainly a collection of user-submitted patches since
the last release.

** New Features

*** Environment variables in a builder (#100)

It is useful to be able to pass environment variables to all steps in a
builder.  This is now possible by adding { .. 'env': { 'var' : 'value' }, ... }
to the builder specification.

*** IRC status plugin improvements (#330, #357, #378, #280, #381, #411, #368)

*** usePTY specified in master.cfg, defaults to False (#158, #255)

Using a pty has some benefits in terms of supporting "Stop Build", but causes
numerous problems with simpler jobs which can be killed by a SIGHUP when their
standard input is closed.  With this change, PTYs are not used by default,
although you can enable them either on slaves (with the --usepty option to
create-slave) or on the master.

*** More information about buildslaves via the web plugin (#110)

A new page, rooted at /buildslave/$SLAVENAME, gives extensive information about
the buildslave.

*** More flexible merging of requests (#415)

The optional c['mergeRequests'] configuration parameter takes a function
which can decide whether two requests are mergeable.

*** Steps can be made to run even if the build has halted (#414)

Adding alwaysRun=True to a step will cause it to run even if some other step
has failed and has haltOnFailure=True.

*** Compress buildstep logfiles (#26)

Logs for each buildstep, which can take a lot of space on a busy buildmaster,
are automatically compressed after the step has finished.

*** Support for "latent" buildslaves

The buildslaves that are started on-demand are called "latent" buildslaves.
Buildbot ships with an abstract base class for building latent buildslaves, and
a concrete implementation for AWS EC2.

*** Customized MailNotifier messages (#175)

MailNotifier now takes an optional function to build the notification message,
allowing ultimate site-level control over the format of buildbot's notification
emails.

*** Nightly scheduler support for building only if changes have occurred

With the addition of onlyIfChanged=True, the Nightly scheduler will not schedule
a new build if no changes have been made since its last scheduled build.

*** Add ATOM/RSS feeds to WebStatus (#372)

Two new pages, /atom and /rss, provide feeds of build events to any feed
reader.  These paths take the same "category" and "branch" arguments as the
waterfall and grid.

*** Add categories to Schedulers and Changes (#182)

This allows a moderate amount of support for multiple projects built in a
single buildmaster.

*** Gracefully shut down a buildslave after its build is complete

The /buildslaves/$SLAVENAME pages have a "Gracefully Shutdown" button which
will cause the corresponding slave to shut itself down when it finishes its
current build.  This is a good way to do work on a slave without causing a
spurious build failure.

*** SVN source steps can send usernames and passwords (#41)

Adding username="foo" and/or password="bar" to an SVN step will cause
--username and --password arguments to be passed to 'svn' on the slave side.
Passwords are suitably obfuscated in logfiles.

** New Steps

*** DirectoryUpload (#393)

This step uploads an entire directory to the master, and can be useful when a
build creates several products (e.g., a client and server package).

*** MasterShelCommand

This step runs a shell command on the server, and can be useful for
post-processing build products, or performing other maintenance tasks on the
master.

*** PyLint (#259)

A PyLint step is available to complement the existing PyFlakes step.

** Bugs Fixed

*** Process output from new versions of Test::Harness (#346)

*** Fixes to the try client and scheduler

*** Remove redundant loop in MailNotifier (#315)

*** Display correct $PWD in logfiles (#179)

*** Do not assume a particular python version on Windows (#401)

*** Sort files in changes (#402)

*** Sort buildslaves lexically (#416)

*** Send properties to all builds initiated by AnyBranchScheduler

*** Dependent Schedulers are more robust to reconfiguration (#35)

*** Fix properties handling in triggered buidls (#392)

*** Use "call" on Windows to avoid errors (#417)

*** Support setDefaultWorkdir in FileUpload and FileDownload (#209)

*** Support WithProperties in FileUpload and FileDownload (#210)

*** Fix a bug where changes could be lost on a master crash (#202)

*** Remove color settings from non-presentation code (#251)

*** Fix builders which stopped working after a PING (#349, #85)

*** Isolate Python exceptions in status plugins (#388)

*** Notify about slaves missing at master startup (#302)

*** Fix tracebacks in web display after a reconfig (#176)

** Version-Control Changes

*** Many Mercurial fixes

 - Inrepo branch support finalized (source step + changegroup hook + test case)
   (#65 #185 #187)

 - Reduced amount of full clones by separating clone with update into
   clone/pull/update steps (#186, #227) (see #412 for future work here)

 - Fixed mercurial changegroup hook to work with Mercurial 1.1 API (#181, #380)

*** Many git fixes

*** Add got_revision to Perforce support (#127)

*** Use "git foo" everywhere instead of deprecated "git-foo"

** Minor Changes

*** factory.addSteps (#317)

If you have a common list of steps that are included in multiple factories, you
can use f.addSteps(steplist) to add them all at once.

*** Twisted logfile rotation and cleanup (#108)

By default, Buildbot now rotates and cleans up the (potentially voluminous)
twistd.log files.

*** Prioritize build requests based on the time they wre submitted (#334)

Balancing of load is a bit more fair, although not true load balancing.


-- 
Storage Software Engineer
http://www.zmanda.com



From macke at yar.nu  Thu Feb 26 20:39:46 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Thu, 26 Feb 2009 21:39:46 +0100
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
Message-ID: <49A6FE12.2060907@yar.nu>

Dustin J. Mitchell wrote:
> I've just finished releasing buildbot-0.7.10, available for download
> in the usual places:

[snip]

> This release has *truly* been a community effort -- I did little more
> than marshall patches from a number of contributors and developers.
> Thank you to everyone who helped to make this release a reality, and
> especially to those who pitched in over the last week or so to quash
> the last few bugs.

You've done an amazing job in pulling this release together! It's not a 
small effort in any way. Be proud!

Personally, it's been really fun to contribute a bit here and there. 
Hopefully the Mercurial & Windows situation has improved a bit from that.

But, I missed the release candidate? Did it fall below my radar or did 
it just not happen? Either way, I'm going to try 0.7.10 at work next 
week and see how it handles. Hopefully no nasty surprises. :)

I'll also see if I can get some time to hack on the py2exe stuff. That's 
time well spent I think, as it helps both with windows adoption and we 
want it for 1.0 (i.e. http://buildbot.net/trac/ticket/273)

Many thanks for keeping the buildbot alive and kicking!

Cheers,
/Marcus




From stephen at starswan.com  Thu Feb 26 20:42:40 2009
From: stephen at starswan.com (Stephen Dicks)
Date: Thu, 26 Feb 2009 20:42:40 +0000
Subject: [Buildbot-devel] The Meta Buildbot
In-Reply-To: <42338fbf0902260845r7c671c42p44c1caa1f2dda782@mail.gmail.com>
References: <9733D02D2B21A8498D7702A721E3036C155DC7B8DC@MFMAIL007.mitrefinch.co.uk>	<49A1A64C.4070401@starswan.com>	<42338fbf0902221609n5d0fe6e3s3e51985e0acc582c@mail.gmail.com>	<77fcfd0a0902260833v2fd9e402w109fd1642014f467@mail.gmail.com>
	<42338fbf0902260845r7c671c42p44c1caa1f2dda782@mail.gmail.com>
Message-ID: <49A6FEC0.8030609@starswan.com>

I will change this to use the other URL.
> There's a bug with pulling via HTTP from github:
>   http://support.github.com/discussions/repos/199-git-repo-missing-a-blob
>
> In the interim, I think that we should be pulling from
>   http://repo.or.cz/r/buildbot.git
> but I haven't heard back from Stephen in a while.
>
> Dustin
>
>   




From giallu at gmail.com  Thu Feb 26 20:53:59 2009
From: giallu at gmail.com (Gianluca Sforna)
Date: Thu, 26 Feb 2009 21:53:59 +0100
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
Message-ID: <d5ad21470902261253v6ec54af0qa618613eb94dcfe2@mail.gmail.com>

On Thu, Feb 26, 2009 at 7:16 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
> I've just finished releasing buildbot-0.7.10, available for download
> in the usual places:

Great! Preparing an update for Fedora...

-- 
Gianluca Sforna

http://morefedora.blogspot.com
http://www.linkedin.com/in/gianlucasforna



From smilner at redhat.com  Thu Feb 26 21:23:28 2009
From: smilner at redhat.com (Steve 'Ashcrow' Milner)
Date: Thu, 26 Feb 2009 16:23:28 -0500
Subject: [Buildbot-devel] Buildbot-0.7.10
In-Reply-To: <d5ad21470902261253v6ec54af0qa618613eb94dcfe2@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<d5ad21470902261253v6ec54af0qa618613eb94dcfe2@mail.gmail.com>
Message-ID: <20090226212328.GR2823@arc.rdu.redhat.com>

On 26/02/09 21:53 +0100, Gianluca Sforna wrote:
>On Thu, Feb 26, 2009 at 7:16 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
>> I've just finished releasing buildbot-0.7.10, available for download
>> in the usual places:
>
>Great! Preparing an update for Fedora...
>
>-- 
>Gianluca Sforna
>
>http://morefedora.blogspot.com
>http://www.linkedin.com/in/gianlucasforna
>
>------------------------------------------------------------------------------
>Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
>-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
>-Strategies to boost innovation and cut costs with open source participation
>-Receive a $600 discount off the registration fee with the source code: SFAD
>http://p.sf.net/sfu/XcvMzF8H
>_______________________________________________
>Buildbot-devel mailing list
>Buildbot-devel at lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/buildbot-devel

Sweet! Great work folks!

Gianluca, I'll help test the package when it hits bodhi.

-- 
kthxbye!
Steve 'Ashcrow' Milner
Agent of Infosec
IRC: ashcrow
GnuPG ID: 28DFD4BE

"In the heat of conversation I may have said certain things I believe 
to be untrue. The alleged lie that you might have heard me saying 
allegedly moments ago ... thats a parasite that lives in my neck." 
     -- Tad Ghostal
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://buildbot.net/pipermail/devel/attachments/20090226/510d5a5f/attachment.bin>

From dustin at zmanda.com  Thu Feb 26 21:25:38 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Thu, 26 Feb 2009 16:25:38 -0500
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <49A6FE12.2060907@yar.nu>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<49A6FE12.2060907@yar.nu>
Message-ID: <42338fbf0902261325y510701fapa18f7e2664bae727@mail.gmail.com>

On Thu, Feb 26, 2009 at 3:39 PM, Marcus Lindblom <macke at yar.nu> wrote:
> But, I missed the release candidate? Did it fall below my radar or did it
> just not happen? Either way, I'm going to try 0.7.10 at work next week and
> see how it handles. Hopefully no nasty surprises. :)

I skipped it -- I'll release a patch if we find any glaring omissions,
but I have deployed this version at Zmanda, and a few others have it
running already, so I'm fairly confident that it's not DOA.

In my experience, betas and release candidates tend not to attract
much attention or testing.  But that may be a situation peculiar to
Amanda :)

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From giallu at gmail.com  Thu Feb 26 21:33:51 2009
From: giallu at gmail.com (Gianluca Sforna)
Date: Thu, 26 Feb 2009 22:33:51 +0100
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <42338fbf0902261325y510701fapa18f7e2664bae727@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<49A6FE12.2060907@yar.nu>
	<42338fbf0902261325y510701fapa18f7e2664bae727@mail.gmail.com>
Message-ID: <d5ad21470902261333k4532918m1b5fbe3627a7f280@mail.gmail.com>

On Thu, Feb 26, 2009 at 10:25 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
>
> In my experience, betas and release candidates tend not to attract
> much attention or testing.  But that may be a situation peculiar to
> Amanda :)
>
Unfortunately that's the rule rather than the exception...

so +1 to release early, release often


-- 
Gianluca Sforna

http://morefedora.blogspot.com
http://www.linkedin.com/in/gianlucasforna



From jpittman at group6.net  Thu Feb 26 23:15:52 2009
From: jpittman at group6.net (Johnnie Pittman)
Date: Thu, 26 Feb 2009 15:15:52 -0800
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <d5ad21470902261333k4532918m1b5fbe3627a7f280@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<49A6FE12.2060907@yar.nu>
	<42338fbf0902261325y510701fapa18f7e2664bae727@mail.gmail.com>
	<d5ad21470902261333k4532918m1b5fbe3627a7f280@mail.gmail.com>
Message-ID: <27a262060902261515p70622bd2w31aa88e04e450e71@mail.gmail.com>

FYI, since the push to 0.7.10, the documentation likes found here no longer
work:

http://buildbot.net/trac/wiki/Documentation



On Thu, Feb 26, 2009 at 1:33 PM, Gianluca Sforna <giallu at gmail.com> wrote:

> On Thu, Feb 26, 2009 at 10:25 PM, Dustin J. Mitchell <dustin at zmanda.com>
> wrote:
> >
> > In my experience, betas and release candidates tend not to attract
> > much attention or testing.  But that may be a situation peculiar to
> > Amanda :)
> >
> Unfortunately that's the rule rather than the exception...
>
> so +1 to release early, release often
>
>
> --
> Gianluca Sforna
>
> http://morefedora.blogspot.com
> http://www.linkedin.com/in/gianlucasforna
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
> CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the
> Enterprise
> -Strategies to boost innovation and cut costs with open source
> participation
> -Receive a $600 discount off the registration fee with the source code:
> SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090226/8eb7eee9/attachment.html>

From mahatmamanic at gmail.com  Fri Feb 27 00:20:27 2009
From: mahatmamanic at gmail.com (Ian Peters-Campbell)
Date: Thu, 26 Feb 2009 16:20:27 -0800
Subject: [Buildbot-devel] split_file
Message-ID: <c02dbb2a0902261620w6272b3a4n150aa312ffb76d0f@mail.gmail.com>

Hey all,

I am about a day and a half from being done with quite the glorious buildbot
installation, and I realized that I had not properly grokked the SVNPoller
split_file.  After reading through the example, I went to write a split file
function, and either I came up with something that is (I think) simpler than
the examples, or alternatively I am way off base with my understanding of
how it works.  If the former then maybe this can help others, and if the
latter maybe others can help me :)

Here is what I wrote for the file splitter:

branches = []

def fileSplitter(path):
    """\
    Returns the branch/file tuples necessary for the poller/schedulers to
    kick off the correct scheduled builds
    """
    for branch in branches:
        if path.startswith(branch):
            return (branch, path.split(branch)[1])

    return None

fileSplitter is then passed as the split_file of the SVNPoller when it is
created.


Then, each time I build a scheduler, I do something like:


scheduler = Scheduler(projectName, branch, stableTimer, builderNames)
branches.append(branch)

Now each time a change is made fileSplitter should look through the list of
branches that apply to my specific projects, and return the proper tuple if
the change occurred in one of them.  If the change occurred outside one of
the project locations I specifically care about, then it is ignored.

Does my aproach here make sense, or am I missing something?  If this follows
it seems like an easier way to parse branch names than splitting and
reforming the strings.

Ian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090226/aa88516e/attachment.html>

From jpittman at group6.net  Thu Feb 26 23:16:24 2009
From: jpittman at group6.net (Johnnie Pittman)
Date: Thu, 26 Feb 2009 15:16:24 -0800
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <27a262060902261515p70622bd2w31aa88e04e450e71@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<49A6FE12.2060907@yar.nu>
	<42338fbf0902261325y510701fapa18f7e2664bae727@mail.gmail.com>
	<d5ad21470902261333k4532918m1b5fbe3627a7f280@mail.gmail.com>
	<27a262060902261515p70622bd2w31aa88e04e450e71@mail.gmail.com>
Message-ID: <27a262060902261516j22b4bbefq844b4649711cade3@mail.gmail.com>

err... links.  not likes... ;)

On Thu, Feb 26, 2009 at 3:15 PM, Johnnie Pittman <jpittman at group6.net>wrote:

> FYI, since the push to 0.7.10, the documentation likes found here no longer
> work:
>
> http://buildbot.net/trac/wiki/Documentation
>
>
>
>
> On Thu, Feb 26, 2009 at 1:33 PM, Gianluca Sforna <giallu at gmail.com> wrote:
>
>> On Thu, Feb 26, 2009 at 10:25 PM, Dustin J. Mitchell <dustin at zmanda.com>
>> wrote:
>> >
>> > In my experience, betas and release candidates tend not to attract
>> > much attention or testing.  But that may be a situation peculiar to
>> > Amanda :)
>> >
>> Unfortunately that's the rule rather than the exception...
>>
>> so +1 to release early, release often
>>
>>
>> --
>> Gianluca Sforna
>>
>> http://morefedora.blogspot.com
>> http://www.linkedin.com/in/gianlucasforna
>>
>>
>> ------------------------------------------------------------------------------
>> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
>> CA
>> -OSBC tackles the biggest issue in open source: Open Sourcing the
>> Enterprise
>> -Strategies to boost innovation and cut costs with open source
>> participation
>> -Receive a $600 discount off the registration fee with the source code:
>> SFAD
>> http://p.sf.net/sfu/XcvMzF8H
>> _______________________________________________
>> Buildbot-devel mailing list
>> Buildbot-devel at lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090226/62f1177f/attachment.html>

From clepple at gmail.com  Fri Feb 27 03:05:26 2009
From: clepple at gmail.com (Charles Lepple)
Date: Thu, 26 Feb 2009 22:05:26 -0500
Subject: [Buildbot-devel] trouble running tests for 0.7.10
Message-ID: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>

Hi all,

Another big thanks to Dustin for collecting all the patches for  
0.7.10 - looks good overall!

I did notice one issue with the tarball, though. It doesn't seem to  
include buildbot.steps.package.rpm, although this was added to the  
Git tree at the same time as test/test_package_rpm.py. Does something  
need to be adjusted in setup.py?

-- 
Charles Lepple
clepple at gmail






From smilner at redhat.com  Fri Feb 27 03:17:55 2009
From: smilner at redhat.com (Steve 'Ashcrow' Milner)
Date: Thu, 26 Feb 2009 22:17:55 -0500
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
Message-ID: <20090227031755.GT2823@arc.rdu.redhat.com>

On 26/02/09 22:05 -0500, Charles Lepple wrote:
>Hi all,
>
>Another big thanks to Dustin for collecting all the patches for  
>0.7.10 - looks good overall!
>
>I did notice one issue with the tarball, though. It doesn't seem to  
>include buildbot.steps.package.rpm, although this was added to the  
>Git tree at the same time as test/test_package_rpm.py. Does something  
>need to be adjusted in setup.py?
>
>-- 
>Charles Lepple
>clepple at gmail
>
>
>
>
>------------------------------------------------------------------------------
>Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
>-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
>-Strategies to boost innovation and cut costs with open source participation
>-Receive a $600 discount off the registration fee with the source code: SFAD
>http://p.sf.net/sfu/XcvMzF8H
>_______________________________________________
>Buildbot-devel mailing list
>Buildbot-devel at lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/buildbot-devel

Yeah, you are right. Sorry, that would be my fault.

This is what I should have added ... I'll commit this and do a pull
request.

$ git diff setup.py
diff --git a/setup.py b/setup.py
index 1a744cd..451f15f 100644
--- a/setup.py
+++ b/setup.py
@@ -84,6 +84,8 @@ setup_args = {
                "buildbot.status", "buildbot.status.web",
                "buildbot.changes",
                "buildbot.steps",
+              "buildbot.steps.package",
+              "buildbot.steps.package.rpm",
                "buildbot.process",
                "buildbot.clients",
                "buildbot.slave",


-- 
kthxbye!
Steve 'Ashcrow' Milner
Agent of Infosec
IRC: ashcrow
GnuPG ID: 28DFD4BE

"In the heat of conversation I may have said certain things I believe 
to be untrue. The alleged lie that you might have heard me saying 
allegedly moments ago ... thats a parasite that lives in my neck." 
     -- Tad Ghostal
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://buildbot.net/pipermail/devel/attachments/20090226/4ea29862/attachment.bin>

From smilner at redhat.com  Fri Feb 27 03:39:08 2009
From: smilner at redhat.com (Steve 'Ashcrow' Milner)
Date: Thu, 26 Feb 2009 22:39:08 -0500
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <20090227031755.GT2823@arc.rdu.redhat.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
	<20090227031755.GT2823@arc.rdu.redhat.com>
Message-ID: <20090227033908.GU2823@arc.rdu.redhat.com>

On 26/02/09 22:17 -0500, Steve 'Ashcrow' Milner wrote:
> On 26/02/09 22:05 -0500, Charles Lepple wrote:
>> Hi all,
>>
>> Another big thanks to Dustin for collecting all the patches for   
>> 0.7.10 - looks good overall!
>>
>> I did notice one issue with the tarball, though. It doesn't seem to   
>> include buildbot.steps.package.rpm, although this was added to the   
>> Git tree at the same time as test/test_package_rpm.py. Does something   
>> need to be adjusted in setup.py?
>>
>> -- 
>> Charles Lepple
>> clepple at gmail
>>
>>
>>
>>
>> ------------------------------------------------------------------------------
>> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
>> -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
>> -Strategies to boost innovation and cut costs with open source participation
>> -Receive a $600 discount off the registration fee with the source code: SFAD
>> http://p.sf.net/sfu/XcvMzF8H
>> _______________________________________________
>> Buildbot-devel mailing list
>> Buildbot-devel at lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
> Yeah, you are right. Sorry, that would be my fault.
>
> This is what I should have added ... I'll commit this and do a pull
> request.
>
> $ git diff setup.py
> diff --git a/setup.py b/setup.py
> index 1a744cd..451f15f 100644
> --- a/setup.py
> +++ b/setup.py
> @@ -84,6 +84,8 @@ setup_args = {
>                "buildbot.status", "buildbot.status.web",
>                "buildbot.changes",
>                "buildbot.steps",
> +              "buildbot.steps.package",
> +              "buildbot.steps.package.rpm",
>                "buildbot.process",
>                "buildbot.clients",
>                "buildbot.slave",
>
>
> -- 
> kthxbye!
> Steve 'Ashcrow' Milner
> Agent of Infosec
> IRC: ashcrow
> GnuPG ID: 28DFD4BE
>
> "In the heat of conversation I may have said certain things I believe to 
> be untrue. The alleged lie that you might have heard me saying allegedly 
> moments ago ... thats a parasite that lives in my neck."     -- Tad 
> Ghostal


After applying the above patch I verified the results with an sdist
built tarball.

buildbot$ trial test
....
Ran 308 tests in 273.265s

PASSED (skips=16, successes=292)

-- 
kthxbye!
Steve 'Ashcrow' Milner
Agent of Infosec
IRC: ashcrow
GnuPG ID: 28DFD4BE

"In the heat of conversation I may have said certain things I believe 
to be untrue. The alleged lie that you might have heard me saying 
allegedly moments ago ... thats a parasite that lives in my neck." 
     -- Tad Ghostal
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://buildbot.net/pipermail/devel/attachments/20090226/92ad8be4/attachment.bin>

From macke at yar.nu  Thu Feb 26 20:39:46 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Thu, 26 Feb 2009 21:39:46 +0100
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
Message-ID: <49A6FE12.2060907@yar.nu>



Dustin J. Mitchell wrote:
> I've just finished releasing buildbot-0.7.10, available for download
> in the usual places:

[snip]

> This release has *truly* been a community effort -- I did little more
> than marshall patches from a number of contributors and developers.
> Thank you to everyone who helped to make this release a reality, and
> especially to those who pitched in over the last week or so to quash
> the last few bugs.

You've done an amazing job in pulling this release together! It's not a 
small effort in any way. Be proud!

Personally, it's been really fun to contribute a bit here and there. 
Hopefully the Mercurial & Windows situation has improved a bit from that.

But, I missed the release candidate? Did it fall below my radar or did 
it just not happen? Either way, I'm going to try 0.7.10 at work next 
week and see how it handles. Hopefully no nasty surprises. :)

I'll also see if I can get some time to hack on the py2exe stuff. That's 
time well spent I think, as it helps both with windows adoption and we 
want it for 1.0 (i.e. http://buildbot.net/trac/ticket/273)

Many thanks for keeping the buildbot alive and kicking!

Cheers,
/Marcus





From giallu at gmail.com  Fri Feb 27 08:08:17 2009
From: giallu at gmail.com (Gianluca Sforna)
Date: Fri, 27 Feb 2009 09:08:17 +0100
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <20090227033908.GU2823@arc.rdu.redhat.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
	<20090227031755.GT2823@arc.rdu.redhat.com>
	<20090227033908.GU2823@arc.rdu.redhat.com>
Message-ID: <d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>

On Fri, Feb 27, 2009 at 4:39 AM, Steve 'Ashcrow' Milner
<smilner at redhat.com> wrote:
>>
>> Yeah, you are right. Sorry, that would be my fault.
>>
>> This is what I should have added ... I'll commit this and do a pull
>> request.
>>
>> $ git diff setup.py
>> diff --git a/setup.py b/setup.py
>> index 1a744cd..451f15f 100644
>> --- a/setup.py
>> +++ b/setup.py
>> @@ -84,6 +84,8 @@ setup_args = {
>> ? ? ? ? ? ? ? "buildbot.status", "buildbot.status.web",
>> ? ? ? ? ? ? ? "buildbot.changes",
>> ? ? ? ? ? ? ? "buildbot.steps",
>> + ? ? ? ? ? ? ?"buildbot.steps.package",
>> + ? ? ? ? ? ? ?"buildbot.steps.package.rpm",
>> ? ? ? ? ? ? ? "buildbot.process",
>> ? ? ? ? ? ? ? "buildbot.clients",
>> ? ? ? ? ? ? ? "buildbot.slave",
>>
>>
>> --

> After applying the above patch I verified the results with an sdist
> built tarball.

I was actually wondering why tests were failing during build...
Does this mean we need new tarball (0.7.11 or something)?



-- 
Gianluca Sforna

http://morefedora.blogspot.com
http://www.linkedin.com/in/gianlucasforna



From macke at yar.nu  Fri Feb 27 08:46:35 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Fri, 27 Feb 2009 09:46:35 +0100
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <42338fbf0902261325y510701fapa18f7e2664bae727@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>	
	<49A6FE12.2060907@yar.nu>
	<42338fbf0902261325y510701fapa18f7e2664bae727@mail.gmail.com>
Message-ID: <49A7A86B.3070500@yar.nu>

Dustin J. Mitchell wrote:
> On Thu, Feb 26, 2009 at 3:39 PM, Marcus Lindblom <macke at yar.nu> wrote:
>   
>> But, I missed the release candidate? Did it fall below my radar or did it
>> just not happen? Either way, I'm going to try 0.7.10 at work next week and
>> see how it handles. Hopefully no nasty surprises. :)
>>     
> I skipped it -- I'll release a patch if we find any glaring omissions,
> but I have deployed this version at Zmanda, and a few others have it
> running already, so I'm fairly confident that it's not DOA.
>
> In my experience, betas and release candidates tend not to attract
> much attention or testing.  But that may be a situation peculiar to
> Amanda :)
>   
You're right about that.  I'm just a bit nervous about the Mercurial 
changes as I haven't run our BB with the new changes, because I was 
waiting for you to finish up the RC.

I guess the time for waiting is over. ;)

Cheers,
/Marcus







From l10n.moz at googlemail.com  Fri Feb 27 11:20:22 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Fri, 27 Feb 2009 12:20:22 +0100
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
Message-ID: <9de1bdb60902270320u562f810dw61008b1b47461900@mail.gmail.com>

Hi,

congrats from me to you and all, too.

In particular the choice for github seems to have worked out really well for
everybody. It's an interesting experience seeing how a bit of web
infrastructure can turn a dvcs into a collaborative effort. I'm pretty sure
it wasn't darcs or git that made the difference here, and, tbh, neither
wether Brian or Dustin drove this release. Watching the network graph just
has been fun all around.

Nitpick, Dustin, did you tag the release? As my local clone doesn't have
that tag.

Axel

2009/2/26 Dustin J. Mitchell <dustin at zmanda.com>

> I've just finished releasing buildbot-0.7.10, available for download
> in the usual places:
>
>  http://pypi.python.org/pypi/buildbot/0.7.10
>  https://sourceforge.net/project/showfiles.php?group_id=73177
>
> File checksums are as follows.  This buildbot release is signed by my
> GPG public key (7F0D15B1) (available from keyservers):
>
>  buildbot-0.7.10.tar.gz
>  md5: 3a51f6ba8a852382863d1940cef89a46
>
>  buildbot-0.7.10.zip
>  md5: 33946b82706e8b65a1f8dab2de7ab9bf
>
> This release has *truly* been a community effort -- I did little more
> than marshall patches from a number of contributors and developers.
> Thank you to everyone who helped to make this release a reality, and
> especially to those who pitched in over the last week or so to quash
> the last few bugs.
>
> ----
>
> This release is mainly a collection of user-submitted patches since
> the last release.
>
> ** New Features
>
> *** Environment variables in a builder (#100)
>
> It is useful to be able to pass environment variables to all steps in a
> builder.  This is now possible by adding { .. 'env': { 'var' : 'value' },
> ... }
> to the builder specification.
>
> *** IRC status plugin improvements (#330, #357, #378, #280, #381, #411,
> #368)
>
> *** usePTY specified in master.cfg, defaults to False (#158, #255)
>
> Using a pty has some benefits in terms of supporting "Stop Build", but
> causes
> numerous problems with simpler jobs which can be killed by a SIGHUP when
> their
> standard input is closed.  With this change, PTYs are not used by default,
> although you can enable them either on slaves (with the --usepty option to
> create-slave) or on the master.
>
> *** More information about buildslaves via the web plugin (#110)
>
> A new page, rooted at /buildslave/$SLAVENAME, gives extensive information
> about
> the buildslave.
>
> *** More flexible merging of requests (#415)
>
> The optional c['mergeRequests'] configuration parameter takes a function
> which can decide whether two requests are mergeable.
>
> *** Steps can be made to run even if the build has halted (#414)
>
> Adding alwaysRun=True to a step will cause it to run even if some other
> step
> has failed and has haltOnFailure=True.
>
> *** Compress buildstep logfiles (#26)
>
> Logs for each buildstep, which can take a lot of space on a busy
> buildmaster,
> are automatically compressed after the step has finished.
>
> *** Support for "latent" buildslaves
>
> The buildslaves that are started on-demand are called "latent" buildslaves.
> Buildbot ships with an abstract base class for building latent buildslaves,
> and
> a concrete implementation for AWS EC2.
>
> *** Customized MailNotifier messages (#175)
>
> MailNotifier now takes an optional function to build the notification
> message,
> allowing ultimate site-level control over the format of buildbot's
> notification
> emails.
>
> *** Nightly scheduler support for building only if changes have occurred
>
> With the addition of onlyIfChanged=True, the Nightly scheduler will not
> schedule
> a new build if no changes have been made since its last scheduled build.
>
> *** Add ATOM/RSS feeds to WebStatus (#372)
>
> Two new pages, /atom and /rss, provide feeds of build events to any feed
> reader.  These paths take the same "category" and "branch" arguments as the
> waterfall and grid.
>
> *** Add categories to Schedulers and Changes (#182)
>
> This allows a moderate amount of support for multiple projects built in a
> single buildmaster.
>
> *** Gracefully shut down a buildslave after its build is complete
>
> The /buildslaves/$SLAVENAME pages have a "Gracefully Shutdown" button which
> will cause the corresponding slave to shut itself down when it finishes its
> current build.  This is a good way to do work on a slave without causing a
> spurious build failure.
>
> *** SVN source steps can send usernames and passwords (#41)
>
> Adding username="foo" and/or password="bar" to an SVN step will cause
> --username and --password arguments to be passed to 'svn' on the slave
> side.
> Passwords are suitably obfuscated in logfiles.
>
> ** New Steps
>
> *** DirectoryUpload (#393)
>
> This step uploads an entire directory to the master, and can be useful when
> a
> build creates several products (e.g., a client and server package).
>
> *** MasterShelCommand
>
> This step runs a shell command on the server, and can be useful for
> post-processing build products, or performing other maintenance tasks on
> the
> master.
>
> *** PyLint (#259)
>
> A PyLint step is available to complement the existing PyFlakes step.
>
> ** Bugs Fixed
>
> *** Process output from new versions of Test::Harness (#346)
>
> *** Fixes to the try client and scheduler
>
> *** Remove redundant loop in MailNotifier (#315)
>
> *** Display correct $PWD in logfiles (#179)
>
> *** Do not assume a particular python version on Windows (#401)
>
> *** Sort files in changes (#402)
>
> *** Sort buildslaves lexically (#416)
>
> *** Send properties to all builds initiated by AnyBranchScheduler
>
> *** Dependent Schedulers are more robust to reconfiguration (#35)
>
> *** Fix properties handling in triggered buidls (#392)
>
> *** Use "call" on Windows to avoid errors (#417)
>
> *** Support setDefaultWorkdir in FileUpload and FileDownload (#209)
>
> *** Support WithProperties in FileUpload and FileDownload (#210)
>
> *** Fix a bug where changes could be lost on a master crash (#202)
>
> *** Remove color settings from non-presentation code (#251)
>
> *** Fix builders which stopped working after a PING (#349, #85)
>
> *** Isolate Python exceptions in status plugins (#388)
>
> *** Notify about slaves missing at master startup (#302)
>
> *** Fix tracebacks in web display after a reconfig (#176)
>
> ** Version-Control Changes
>
> *** Many Mercurial fixes
>
>  - Inrepo branch support finalized (source step + changegroup hook + test
> case)
>   (#65 #185 #187)
>
>  - Reduced amount of full clones by separating clone with update into
>   clone/pull/update steps (#186, #227) (see #412 for future work here)
>
>  - Fixed mercurial changegroup hook to work with Mercurial 1.1 API (#181,
> #380)
>
> *** Many git fixes
>
> *** Add got_revision to Perforce support (#127)
>
> *** Use "git foo" everywhere instead of deprecated "git-foo"
>
> ** Minor Changes
>
> *** factory.addSteps (#317)
>
> If you have a common list of steps that are included in multiple factories,
> you
> can use f.addSteps(steplist) to add them all at once.
>
> *** Twisted logfile rotation and cleanup (#108)
>
> By default, Buildbot now rotates and cleans up the (potentially voluminous)
> twistd.log files.
>
> *** Prioritize build requests based on the time they wre submitted (#334)
>
> Balancing of load is a bit more fair, although not true load balancing.
>
>
> --
> Storage Software Engineer
> http://www.zmanda.com
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
> CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the
> Enterprise
> -Strategies to boost innovation and cut costs with open source
> participation
> -Receive a $600 discount off the registration fee with the source code:
> SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090227/455f084f/attachment.html>

From giallu at gmail.com  Fri Feb 27 11:50:54 2009
From: giallu at gmail.com (Gianluca Sforna)
Date: Fri, 27 Feb 2009 12:50:54 +0100
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <9de1bdb60902270320u562f810dw61008b1b47461900@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<9de1bdb60902270320u562f810dw61008b1b47461900@mail.gmail.com>
Message-ID: <d5ad21470902270350m1ad81e83j6f7a88149536e598@mail.gmail.com>

On Fri, Feb 27, 2009 at 12:20 PM, Axel Hecht <l10n.moz at googlemail.com> wrote:
>
> Nitpick, Dustin, did you tag the release? As my local clone doesn't have
> that tag.
>

I see the tag from the github interface:
http://github.com/djmitche/buildbot/tree/buildbot-0.7.10

IIRC your need to pass an option to 'git fetch' to download the remote tags.

-- 
Gianluca Sforna

http://morefedora.blogspot.com
http://www.linkedin.com/in/gianlucasforna



From cdevienne at gmail.com  Fri Feb 27 12:04:11 2009
From: cdevienne at gmail.com (Christophe de VIENNE)
Date: Fri, 27 Feb 2009 13:04:11 +0100
Subject: [Buildbot-devel] Mercurial revision on named branches
Message-ID: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>

Hi,

I am having a very annoying issue with buildbot / mercurial that makes
it impossible to use tags for building a precise version of our
software.

I have created a trac ticket describing the issue more precisely :

http://buildbot.net/trac/ticket/438

I also see that the bug tracker on sourceforge is active : do I need
to post the issue there instead of trac ?

Thanks,

Christophe



From l10n.moz at googlemail.com  Fri Feb 27 12:56:07 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Fri, 27 Feb 2009 13:56:07 +0100
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <d5ad21470902270350m1ad81e83j6f7a88149536e598@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<9de1bdb60902270320u562f810dw61008b1b47461900@mail.gmail.com>
	<d5ad21470902270350m1ad81e83j6f7a88149536e598@mail.gmail.com>
Message-ID: <9de1bdb60902270456h42969781t2c81faeeb4ce1b5d@mail.gmail.com>

Ah, thanks.

git pull -t dustin master

helped.

Axel

2009/2/27 Gianluca Sforna <giallu at gmail.com>

> On Fri, Feb 27, 2009 at 12:20 PM, Axel Hecht <l10n.moz at googlemail.com>
> wrote:
> >
> > Nitpick, Dustin, did you tag the release? As my local clone doesn't have
> > that tag.
> >
>
> I see the tag from the github interface:
> http://github.com/djmitche/buildbot/tree/buildbot-0.7.10
>
> IIRC your need to pass an option to 'git fetch' to download the remote
> tags.
>
> --
> Gianluca Sforna
>
> http://morefedora.blogspot.com
> http://www.linkedin.com/in/gianlucasforna
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco,
> CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the
> Enterprise
> -Strategies to boost innovation and cut costs with open source
> participation
> -Receive a $600 discount off the registration fee with the source code:
> SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090227/02e1421e/attachment.html>

From gerg.ward+buildbot at gmail.com  Fri Feb 27 13:14:37 2009
From: gerg.ward+buildbot at gmail.com (Greg Ward)
Date: Fri, 27 Feb 2009 08:14:37 -0500
Subject: [Buildbot-devel] Mercurial revision on named branches
In-Reply-To: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>
References: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>
Message-ID: <cd4419280902270514p6b3d6e44w488c704d0fe1d116@mail.gmail.com>

On Fri, Feb 27, 2009 at 7:04 AM, Christophe de VIENNE
<cdevienne at gmail.com> wrote:
> I also see that the bug tracker on sourceforge is active : do I need
> to post the issue there instead of trac ?

I believe that the SF bug tracker is no longer used for BuildBot;  we
should only be using trac.

Greg



From l10n.moz at googlemail.com  Fri Feb 27 13:17:58 2009
From: l10n.moz at googlemail.com (Axel Hecht)
Date: Fri, 27 Feb 2009 14:17:58 +0100
Subject: [Buildbot-devel] Mercurial revision on named branches
In-Reply-To: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>
References: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>
Message-ID: <9de1bdb60902270517u20570f6y5740d55caf9520cb@mail.gmail.com>

2009/2/27 Christophe de VIENNE <cdevienne at gmail.com>

> Hi,
>
> I am having a very annoying issue with buildbot / mercurial that makes
> it impossible to use tags for building a precise version of our
> software.
>
> I have created a trac ticket describing the issue more precisely :
>
> http://buildbot.net/trac/ticket/438
>
> I also see that the bug tracker on sourceforge is active : do I need
> to post the issue there instead of trac ?
>

Trac-only, please.

Commented in the ticket, and CCed Marcus, who rewrote the hg steps for .10.

Axel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://buildbot.net/pipermail/devel/attachments/20090227/5a2d9705/attachment.html>

From smilner at redhat.com  Fri Feb 27 14:30:45 2009
From: smilner at redhat.com (Steve 'Ashcrow' Milner)
Date: Fri, 27 Feb 2009 09:30:45 -0500
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
	<20090227031755.GT2823@arc.rdu.redhat.com>
	<20090227033908.GU2823@arc.rdu.redhat.com>
	<d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>
Message-ID: <20090227143045.GV2823@arc.rdu.redhat.com>

On 27/02/09 09:08 +0100, Gianluca Sforna wrote:
>On Fri, Feb 27, 2009 at 4:39 AM, Steve 'Ashcrow' Milner
><smilner at redhat.com> wrote:
>>>
>>> Yeah, you are right. Sorry, that would be my fault.
>>>
>>> This is what I should have added ... I'll commit this and do a pull
>>> request.
>>>
>>> $ git diff setup.py
>>> diff --git a/setup.py b/setup.py
>>> index 1a744cd..451f15f 100644
>>> --- a/setup.py
>>> +++ b/setup.py
>>> @@ -84,6 +84,8 @@ setup_args = {
>>> ? ? ? ? ? ? ? "buildbot.status", "buildbot.status.web",
>>> ? ? ? ? ? ? ? "buildbot.changes",
>>> ? ? ? ? ? ? ? "buildbot.steps",
>>> + ? ? ? ? ? ? ?"buildbot.steps.package",
>>> + ? ? ? ? ? ? ?"buildbot.steps.package.rpm",
>>> ? ? ? ? ? ? ? "buildbot.process",
>>> ? ? ? ? ? ? ? "buildbot.clients",
>>> ? ? ? ? ? ? ? "buildbot.slave",
>>>
>>>
>>> --
>
>> After applying the above patch I verified the results with an sdist
>> built tarball.
>
>I was actually wondering why tests were failing during build...
>Does this mean we need new tarball (0.7.11 or something)?
>
>
>
>-- 
>Gianluca Sforna
>
>http://morefedora.blogspot.com
>http://www.linkedin.com/in/gianlucasforna

There are multiple ways to take care of this ... one way is to release
a 0.7.11, another a 0.7.10.1 or lastly it can be done with patch
levels in packages (like %patch0 in rpm). If it isn't a problem for
people it can also be left like it is and fixed in NEXTRELEASE
just noting in the release notes what must be done if you want the
package.* steps.

Again, sorry guys, this was my gaffe.

-- 
kthxbye!
Steve 'Ashcrow' Milner
Agent of Infosec
IRC: ashcrow
GnuPG ID: 28DFD4BE

"In the heat of conversation I may have said certain things I believe 
to be untrue. The alleged lie that you might have heard me saying 
allegedly moments ago ... thats a parasite that lives in my neck." 
     -- Tad Ghostal
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://buildbot.net/pipermail/devel/attachments/20090227/26d092a4/attachment.bin>

From clepple at gmail.com  Fri Feb 27 15:34:53 2009
From: clepple at gmail.com (Charles Lepple)
Date: Fri, 27 Feb 2009 10:34:53 -0500
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <20090227143045.GV2823@arc.rdu.redhat.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
	<20090227031755.GT2823@arc.rdu.redhat.com>
	<20090227033908.GU2823@arc.rdu.redhat.com>
	<d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>
	<20090227143045.GV2823@arc.rdu.redhat.com>
Message-ID: <af8966460902270734o71915c99ndce886b7320675f9@mail.gmail.com>

On Fri, Feb 27, 2009 at 9:30 AM, Steve 'Ashcrow' Milner
<smilner at redhat.com> wrote:
> On 27/02/09 09:08 +0100, Gianluca Sforna wrote:
>>
>> On Fri, Feb 27, 2009 at 4:39 AM, Steve 'Ashcrow' Milner
>> <smilner at redhat.com> wrote:
>>>>
>>>> Yeah, you are right. Sorry, that would be my fault.
>>>>
>>>> This is what I should have added ... I'll commit this and do a pull
>>>> request.
>>>>
>>>> $ git diff setup.py
>>>> diff --git a/setup.py b/setup.py
>>>> index 1a744cd..451f15f 100644
>>>> --- a/setup.py
>>>> +++ b/setup.py
>>>> @@ -84,6 +84,8 @@ setup_args = {
>>>> ? ? ? ? ? ? ? "buildbot.status", "buildbot.status.web",
>>>> ? ? ? ? ? ? ? "buildbot.changes",
>>>> ? ? ? ? ? ? ? "buildbot.steps",
>>>> + ? ? ? ? ? ? ?"buildbot.steps.package",
>>>> + ? ? ? ? ? ? ?"buildbot.steps.package.rpm",
>>>> ? ? ? ? ? ? ? "buildbot.process",
>>>> ? ? ? ? ? ? ? "buildbot.clients",
>>>> ? ? ? ? ? ? ? "buildbot.slave",
>>>>
>>>>
>>>> --
>>
>>> After applying the above patch I verified the results with an sdist
>>> built tarball.
>>
>> I was actually wondering why tests were failing during build...
>> Does this mean we need new tarball (0.7.11 or something)?
>>
>>
>>
>> --
>> Gianluca Sforna
>>
>> http://morefedora.blogspot.com
>> http://www.linkedin.com/in/gianlucasforna
>
> There are multiple ways to take care of this ... one way is to release
> a 0.7.11, another a 0.7.10.1 or lastly it can be done with patch
> levels in packages (like %patch0 in rpm). If it isn't a problem for
> people it can also be left like it is and fixed in NEXTRELEASE
> just noting in the release notes what must be done if you want the
> package.* steps.
>
> Again, sorry guys, this was my gaffe.

No worries on my end - I was actually updating the Fink package for
Mac OS X, which doesn't require RPM support anyway. I just wanted to
let you know.

-- 
- Charles Lepple



From dustin at zmanda.com  Fri Feb 27 15:41:11 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Fri, 27 Feb 2009 10:41:11 -0500
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <20090227143045.GV2823@arc.rdu.redhat.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
	<20090227031755.GT2823@arc.rdu.redhat.com>
	<20090227033908.GU2823@arc.rdu.redhat.com>
	<d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>
	<20090227143045.GV2823@arc.rdu.redhat.com>
Message-ID: <42338fbf0902270741u350aa87cpd07d99a2f03046aa@mail.gmail.com>

On Fri, Feb 27, 2009 at 9:30 AM, Steve 'Ashcrow' Milner
<smilner at redhat.com> wrote:
> There are multiple ways to take care of this ... one way is to release
> a 0.7.11, another a 0.7.10.1 or lastly it can be done with patch
> levels in packages (like %patch0 in rpm). If it isn't a problem for
> people it can also be left like it is and fixed in NEXTRELEASE
> just noting in the release notes what must be done if you want the
> package.* steps.

I'll do a 0.7.10.1, since Steve worked hard on those steps and they
really should be included.  Let's give it a day or two to settle out
any other bugs.

By the way, the position of 0.7.11 bug-wrangler is available, if
anyone wants it.  I'll be active, and I'm happy to cut the actual
release, but for the next 2mo or so I won't be paying much attention
to the bugs in Trac, so I can concentrate on one-oh.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Fri Feb 27 15:45:04 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Fri, 27 Feb 2009 10:45:04 -0500
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <27a262060902261515p70622bd2w31aa88e04e450e71@mail.gmail.com>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<49A6FE12.2060907@yar.nu>
	<42338fbf0902261325y510701fapa18f7e2664bae727@mail.gmail.com>
	<d5ad21470902261333k4532918m1b5fbe3627a7f280@mail.gmail.com>
	<27a262060902261515p70622bd2w31aa88e04e450e71@mail.gmail.com>
Message-ID: <42338fbf0902270745l5b4f2886sa72d5424d378d39a@mail.gmail.com>

On Thu, Feb 26, 2009 at 6:15 PM, Johnnie Pittman <jpittman at group6.net> wrote:
> FYI, since the push to 0.7.10, the documentation likes found here no longer
> work:
>
> http://buildbot.net/trac/wiki/Documentation

I've since updated the documentation.  I'm hosting docs (and any other
web content) using github's "Pages", which means they're in the
gh-pages branch of the same repository:
  http://github.com/djmitche/buildbot/tree/gh-pages

So we can use the same github process (fork, change, pull request) to
update the docs.

I'm mentally toying with the idea of moving the info currently on the
wiki over to the github pages, too, since the wiki doesn't seem to be
acting as a collaboration site.  And since it's full of hidden spam
links.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From dustin at zmanda.com  Fri Feb 27 15:46:24 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Fri, 27 Feb 2009 10:46:24 -0500
Subject: [Buildbot-devel] Mercurial revision on named branches
In-Reply-To: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>
References: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>
Message-ID: <42338fbf0902270746m49189430r6c2bd902e18fba2c@mail.gmail.com>

On Fri, Feb 27, 2009 at 7:04 AM, Christophe de VIENNE
<cdevienne at gmail.com> wrote:
> I also see that the bug tracker on sourceforge is active : do I need
> to post the issue there instead of trac ?

Right near the top of my TODO list is "clean out tickets on SF
trackers and hide the trackers" so that others aren't confused by the
same thing.  I just did this for Amanda last night :)

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From deejay1 at srem.org  Fri Feb 27 21:03:12 2009
From: deejay1 at srem.org (=?UTF-8?B?xYF1a2FzeiBKZXJuYcWb?=)
Date: Fri, 27 Feb 2009 22:03:12 +0100
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <42338fbf0902270741u350aa87cpd07d99a2f03046aa@mail.gmail.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
	<20090227031755.GT2823@arc.rdu.redhat.com>
	<20090227033908.GU2823@arc.rdu.redhat.com>
	<d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>
	<20090227143045.GV2823@arc.rdu.redhat.com>
	<42338fbf0902270741u350aa87cpd07d99a2f03046aa@mail.gmail.com>
Message-ID: <77fcfd0a0902271303x21d16febq91877e145b8e4176@mail.gmail.com>

Hi.

Maybe a bit off topic, but is some looking into the Mercurial test
suite? It still fails on my builder...
I don't know what should I look for, so I found only this in the logs:

2009/02/27 23:06 +0200 [-] --> buildbot.test.test_vc.Mercurial.testCheckout <--
2009/02/27 23:06 +0200 [-] command ['/usr/bin/hg', 'commit', '-m',
'initial_import'] finished with exit code 1
2009/02/27 23:06 +0200 [-]  and stdout
2009/02/27 23:06 +0200 [-]  and stderr Traceback (most recent call last):
          File "/usr/bin/hg", line 20, in <module>
            mercurial.dispatch.run()
          File "/var/lib/python-support/python2.5/mercurial/dispatch.py",
line 20, in run
            sys.exit(dispatch(sys.argv[1:]))
          File "/var/lib/python-support/python2.5/mercurial/dispatch.py",
line 29, in dispatch
            return _runcatch(u, args)
          File "/var/lib/python-support/python2.5/mercurial/dispatch.py",
line 93, in _runcatch
            ui.warn(_("abort: %s\n") % inst[1])
        IndexError: tuple index out of range


Regards,
-- 
?ukasz



From dustin at zmanda.com  Fri Feb 27 21:16:36 2009
From: dustin at zmanda.com (Dustin J. Mitchell)
Date: Fri, 27 Feb 2009 16:16:36 -0500
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <77fcfd0a0902271303x21d16febq91877e145b8e4176@mail.gmail.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
	<20090227031755.GT2823@arc.rdu.redhat.com>
	<20090227033908.GU2823@arc.rdu.redhat.com>
	<d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>
	<20090227143045.GV2823@arc.rdu.redhat.com>
	<42338fbf0902270741u350aa87cpd07d99a2f03046aa@mail.gmail.com>
	<77fcfd0a0902271303x21d16febq91877e145b8e4176@mail.gmail.com>
Message-ID: <42338fbf0902271316w4f7eec8bg4d8034e2da7f7227@mail.gmail.com>

As far as I can tell, that is a traceback of /usr/bin/hg, not
buildbot, so something funny is going on with the hg executable there.
 Is there a .hgrc that might be causing problems?  Can you run those
commands successfully from the command-line?

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com



From macke at yar.nu  Fri Feb 27 21:28:33 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Fri, 27 Feb 2009 22:28:33 +0100
Subject: [Buildbot-devel] Mercurial revision on named branches
In-Reply-To: <9de1bdb60902270517u20570f6y5740d55caf9520cb@mail.gmail.com>
References: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>
	<9de1bdb60902270517u20570f6y5740d55caf9520cb@mail.gmail.com>
Message-ID: <49A85B01.5000805@yar.nu>

Axel Hecht wrote:
> 2009/2/27 Christophe de VIENNE 
> <cdevienne-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org 
> <mailto:cdevienne-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org>>
> 
>     Hi,
> 
>     I am having a very annoying issue with buildbot / mercurial that makes
>     it impossible to use tags for building a precise version of our
>     software.
> 
>     I have created a trac ticket describing the issue more precisely :
> 
>     http://buildbot.net/trac/ticket/438
> 
> Commented in the ticket, and CCed Marcus, who rewrote the hg steps for .10.

FYI for the list: Closed as an EBKAC. ;) (Wasn't using 0.7.10 on client)

Nice to hear that it's working. :)

Cheers,
/Marcus



From macke at yar.nu  Fri Feb 27 21:28:33 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Fri, 27 Feb 2009 22:28:33 +0100
Subject: [Buildbot-devel] Mercurial revision on named branches
In-Reply-To: <9de1bdb60902270517u20570f6y5740d55caf9520cb@mail.gmail.com>
References: <57b32ba80902270404y6a111d65o9daa9a9ab42391f5@mail.gmail.com>
	<9de1bdb60902270517u20570f6y5740d55caf9520cb@mail.gmail.com>
Message-ID: <49A85B01.5000805@yar.nu>

Axel Hecht wrote:
> 2009/2/27 Christophe de VIENNE 
> <cdevienne at gmail.com 
> <mailto:cdevienne at gmail.com>>
> 
>     Hi,
> 
>     I am having a very annoying issue with buildbot / mercurial that makes
>     it impossible to use tags for building a precise version of our
>     software.
> 
>     I have created a trac ticket describing the issue more precisely :
> 
>     http://buildbot.net/trac/ticket/438
> 
> Commented in the ticket, and CCed Marcus, who rewrote the hg steps for .10.

FYI for the list: Closed as an EBKAC. ;) (Wasn't using 0.7.10 on client)

Nice to hear that it's working. :)

Cheers,
/Marcus




From joduinn at mozilla.com  Fri Feb 27 23:51:30 2009
From: joduinn at mozilla.com (John O'Duinn)
Date: Fri, 27 Feb 2009 15:51:30 -0800
Subject: [Buildbot-devel] [ANN] Buildbot-0.7.10
In-Reply-To: <49A6FE12.2060907@yar.nu>
References: <42338fbf0902261016r3437938avff50f067e173a325@mail.gmail.com>
	<49A6FE12.2060907@yar.nu>
Message-ID: <49A87C82.2010608@mozilla.com>

hi Dustin;

Marcus Lindblom wrote:
> Dustin J. Mitchell wrote:
>> I've just finished releasing buildbot-0.7.10, available for download
>> in the usual places:
> 
> [snip]
> 
>> This release has *truly* been a community effort -- I did little more
>> than marshall patches from a number of contributors and developers.
>> Thank you to everyone who helped to make this release a reality, and
>> especially to those who pitched in over the last week or so to quash
>> the last few bugs.
> 
> You've done an amazing job in pulling this release together! It's not a 
> small effort in any way. Be proud!
[snip]
> Many thanks for keeping the buildbot alive and kicking!
> Cheers,
> /Marcus


I have to echo what Marcus said.

Doing a release is complex, and you did it all smoothly. More amazingly,
people should keep in mind this was Dustin's *FIRST* time driving a
BuildBot release... and it just rocked!

All I can say is: wow - nice work, Dustin.


Thank you.
John.



From deejay1 at srem.org  Sat Feb 28 11:54:46 2009
From: deejay1 at srem.org (=?UTF-8?B?xYF1a2FzeiBKZXJuYcWb?=)
Date: Sat, 28 Feb 2009 12:54:46 +0100
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <42338fbf0902271316w4f7eec8bg4d8034e2da7f7227@mail.gmail.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>
	<20090227031755.GT2823@arc.rdu.redhat.com>
	<20090227033908.GU2823@arc.rdu.redhat.com>
	<d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>
	<20090227143045.GV2823@arc.rdu.redhat.com>
	<42338fbf0902270741u350aa87cpd07d99a2f03046aa@mail.gmail.com>
	<77fcfd0a0902271303x21d16febq91877e145b8e4176@mail.gmail.com>
	<42338fbf0902271316w4f7eec8bg4d8034e2da7f7227@mail.gmail.com>
Message-ID: <77fcfd0a0902280354o4cf4afe6p83a6c6b98fdfc6a0@mail.gmail.com>

On Fri, Feb 27, 2009 at 10:16 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
> As far as I can tell, that is a traceback of /usr/bin/hg, not
> buildbot, so something funny is going on with the hg executable there.
> ?Is there a .hgrc that might be causing problems? ?Can you run those
> commands successfully from the command-line?

Hmm, there is no .hgrc on that account and running some hg commands
from the command line seems to work fine. No other buildslaves seem to
have mercurial installed so I can't tell if it's only me...

-- 
?ukasz



From macke at yar.nu  Sat Feb 28 18:18:26 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Sat, 28 Feb 2009 19:18:26 +0100
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <77fcfd0a0902280354o4cf4afe6p83a6c6b98fdfc6a0@mail.gmail.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>	<20090227031755.GT2823@arc.rdu.redhat.com>	<20090227033908.GU2823@arc.rdu.redhat.com>	<d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>	<20090227143045.GV2823@arc.rdu.redhat.com>	<42338fbf0902270741u350aa87cpd07d99a2f03046aa@mail.gmail.com>	<77fcfd0a0902271303x21d16febq91877e145b8e4176@mail.gmail.com>	<42338fbf0902271316w4f7eec8bg4d8034e2da7f7227@mail.gmail.com>
	<77fcfd0a0902280354o4cf4afe6p83a6c6b98fdfc6a0@mail.gmail.com>
Message-ID: <49A97FF2.40805@yar.nu>

?ukasz Jerna? wrote:
> On Fri, Feb 27, 2009 at 10:16 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
>> As far as I can tell, that is a traceback of /usr/bin/hg, not
>> buildbot, so something funny is going on with the hg executable there.
>>  Is there a .hgrc that might be causing problems?  Can you run those
>> commands successfully from the command-line?
> 
> Hmm, there is no .hgrc on that account and running some hg commands
> from the command line seems to work fine. No other buildslaves seem to
> have mercurial installed so I can't tell if it's only me...

There is also a system-wide hgrc in /etc/mercurial on unixy systems (at 
least, I've got one in cygwin.)

Which Mercurial version are you running?

Also, can you run the test-suite locally on that slave and see what 
happens? (i.e. 'trial.py buildbot.test.test_vc.Mercurial')

Cheers,
/Marcus





From macke at yar.nu  Sat Feb 28 18:18:26 2009
From: macke at yar.nu (Marcus Lindblom)
Date: Sat, 28 Feb 2009 19:18:26 +0100
Subject: [Buildbot-devel] trouble running tests for 0.7.10
In-Reply-To: <77fcfd0a0902280354o4cf4afe6p83a6c6b98fdfc6a0@mail.gmail.com>
References: <2C27491A-F72F-4E1A-AA5B-49A0A66B3AD6@gmail.com>	<20090227031755.GT2823@arc.rdu.redhat.com>	<20090227033908.GU2823@arc.rdu.redhat.com>	<d5ad21470902270008x7ada4ac6g8291718eda42df38@mail.gmail.com>	<20090227143045.GV2823@arc.rdu.redhat.com>	<42338fbf0902270741u350aa87cpd07d99a2f03046aa@mail.gmail.com>	<77fcfd0a0902271303x21d16febq91877e145b8e4176@mail.gmail.com>	<42338fbf0902271316w4f7eec8bg4d8034e2da7f7227@mail.gmail.com>
	<77fcfd0a0902280354o4cf4afe6p83a6c6b98fdfc6a0@mail.gmail.com>
Message-ID: <49A97FF2.40805@yar.nu>

?ukasz Jerna? wrote:
> On Fri, Feb 27, 2009 at 10:16 PM, Dustin J. Mitchell <dustin at zmanda.com> wrote:
>> As far as I can tell, that is a traceback of /usr/bin/hg, not
>> buildbot, so something funny is going on with the hg executable there.
>>  Is there a .hgrc that might be causing problems?  Can you run those
>> commands successfully from the command-line?
> 
> Hmm, there is no .hgrc on that account and running some hg commands
> from the command line seems to work fine. No other buildslaves seem to
> have mercurial installed so I can't tell if it's only me...

There is also a system-wide hgrc in /etc/mercurial on unixy systems (at 
least, I've got one in cygwin.)

Which Mercurial version are you running?

Also, can you run the test-suite locally on that slave and see what 
happens? (i.e. 'trial.py buildbot.test.test_vc.Mercurial')

Cheers,
/Marcus