[Buildbot-devel] mercurial builds

[Oleg Smolsky]

Hey Georges,

On 2012-07-31 03:38, Georges Racinet wrote:
>
>
>>
>> Here is the change coming in:
>> 2012-07-30 19:30:21-0700 [Broker,3,127.0.0.1] perspective_addChange 
>> called
>> 2012-07-30 19:30:21-0700 [Broker,3,127.0.0.1] checking for User 
>> Object from hg Change for: Oleg Smolsky <oleg at xxxxx.xxx>
>> 2012-07-30 19:30:21-0700 [-] added change 
>> Change(revision=u'1a2242c644fcf08728ee5524af0c6be9a03b6c5d', 
>> who=u'Oleg  Smolsky <oleg at xxxxx.xxx>', branch=u'default', 
>> comments=u"cblock: fixed an 'order' warning", when=1343701821, 
>> category=None, project=u'', repository=u'http://hg.lan.colo/core') to 
>> database
>>
>> A new build is initiated but the code dies in an assertion:
>> 2012-07-30 19:31:22-0700 [-] BuildStep.failed; traceback follows
>>         Traceback (most recent call last):
>>           File 
>> "/opt/buildbot/sandbox/local/lib/python2.7/site-packages/buildbot-0.8.6p1-py2.7.egg/buildbot/process/buildstep.py", 
>> line 545, in _startStep_2
>>             doStep.addCallback(self._startStep_3)
>>           File 
>> "/opt/buildbot/sandbox/local/lib/python2.7/site-packages/Twisted-12.1.0-py2.7-linux-x86_64.egg/twisted/internet/defer.py", 
>> line 301, in addCallback
>>             callbackKeywords=kw)
>>           File 
>> "/opt/buildbot/sandbox/local/lib/python2.7/site-packages/Twisted-12.1.0-py2.7-linux-x86_64.egg/twisted/internet/defer.py", 
>> line 290, in addCallbacks
>>             self._runCallbacks()
>>           File 
>> "/opt/buildbot/sandbox/local/lib/python2.7/site-packages/Twisted-12.1.0-py2.7-linux-x86_64.egg/twisted/internet/defer.py", 
>> line 551, in _runCallbacks
>>             current.result = callback(current.result, *args, **kw)
>>         --- <exception caught here> ---
>>           File 
>> "/opt/buildbot/sandbox/local/lib/python2.7/site-packages/buildbot-0.8.6p1-py2.7.egg/buildbot/process/buildstep.py", 
>> line 551, in _startStep_3
>>             if self.start() == SKIPPED:
>>           File 
>> "/opt/buildbot/sandbox/local/lib/python2.7/site-packages/buildbot-0.8.6p1-py2.7.egg/buildbot/steps/source/oldsource.py", 
>> line 291, in start
>>             self.startVC(branch, revision, patch)
>>           File 
>> "/opt/buildbot/sandbox/local/lib/python2.7/site-packages/buildbot-0.8.6p1-py2.7.egg/buildbot/steps/source/oldsource.py", 
>> line 1206, in startVC
>>             assert self.branchType == 'inrepo' or not branch
>>         exceptions.AssertionError:
>
> Are you using the older "slave-side" Mercurial step 
> (http://buildbot.net/buildbot/docs/latest/manual/cfg-buildsteps.html#module-buildbot.steps.source, 
> "Caution" paragraph)
Indeed, I was. Just upgraded to the new goodness.

>
>>
>> I have "branchtype" set to "inrepo", but that makes no difference - 
>> the branch is correctly set to "default" anyway...
>> [hgbuildbot]
>> master = localhost:8000
>> branchtype = inrepo
>
> That's the hook config. I'd guess you set branch='default' in the 
> Mercurial step definition, and forgot to add branchtype='inrepo'
Yes, that is correct. I didn't know about that additional spec.

>
>>
>> So, I hacked the assertion out and the build actually succeeds:
>> root at gate:/opt/buildbot/sandbox/lib/python2.7/site-packages/buildbot-0.8.6p1-py2.7.egg/buildbot/steps/source# 
>> diff oldsource.py.orig oldsource.py
>> 1206c1206
>> <             assert self.branchType == 'inrepo' or not branch
>> ---
>> >             #assert self.branchType == 'inrepo' or not branch
>> 1208,1209c1208,1209
>> <             if branch:
>> <                 self.args['branch'] = branch
>> ---
>> >             #if branch:
>> >             #    self.args['branch'] = branch
>>
>> Could somebody clarify how this is really supposed to work please?
>
> That would probably not have worked with another branch than 'default'.
> Here's a working config with the master-side source step, on 0.8.6p1 :
>
> from buildbot.steps.source.mercurial import Mercurial
> factory.addStep(Mercurial(
>     repourl= "https://example.com/repos/core"
>     mode='full',
>     method='fresh',
>     branchType='inrepo',
>     haltOnFailure=True,
>     description='hg:core',
>     workdir='build/calculs.core',
>     ))
Ah, I see. I was missing the "branchType" piece. This change alone fixed 
the issue and I was able to undo the hack. The newer Mercurial piece is 
nicer too - it logs clearly what is being pull'd and how the working dir 
is updated.

Thank you very much for the quick response!

Oleg.