[Buildbot-devel] slave exception in 0.5 & CVS

[Timothee Besset]

This is the latest CVS, both master and slave. commands.py line 324 is
checking args['dir'], which is present but set to none. It's not happening
on my Linux slave, didn't check to see if that code path was not being
hit, if the input data is right there.

TTimo

On Sun, 29 Aug 2004 13:41:38 -0700 (PDT)
Brian Warner <warner-buildbot at lothar.com> wrote:

> From: Timothee Besset <ttimo at idsoftware.com>
> > 
> > I am getting this exception with a test slave running on OSX panther.
> > 
> > Basically the 'dir' key is present, but the value is None. I fixed this in
> > my local source with an additional check. Might want to fix this in CVS?
> 
> Which version were you running? buildbot.slave.commands wasn't split into a
> separate module until post-0.5.0 . What's the cvs version of your
> commands.py?
> 
> In r1.3, I refactored the CVS and SVN slavecommands into a common base class,
> which accepts a ['workdir'] argument as the preferred way to express where
> the buildable sources should end up. This should replace the 'dir' argument,
> which (I think) was used by the SVN step to express where the SVN checkout
> should take place (which is not exactly the same concept).
> 
> The new "checkout mode" argument should simplify this stuff a lot, but it
> will definitely require a sanity check before the next release to try to deal
> with old slaves that don't take the same arguments. I've started to add a
> version-reporting mechanism, so the master can find out what version of each
> remote command is available (and taylor its 'args' dict to match), so maybe
> that will provide the answer.
> 
> In short, could you check current CVS and see if the problem is still
> present? (and make sure you report the CVS revision of the file, since this
> stuff has changed pretty quickly in the last week).
> 
> thanks,
>  -Brian
>