[Twisted-Python] Twisted 10.1.0 release status update
Hello everyone, The Twisted 10.1.0 release is now a month overdue. Sorry. We've got a pre-release which appears to be free of critical bugs, and I'm up to step 4 of "How to do a final release" on http://twistedmatrix.com/trac/wiki/ReleaseProcess. My intent is to complete the final steps of the release this weekend. It would be very, very helpful to have someone around who has root on cube and is able to be responsive to requests during a UK afternoon / evening. Twisted prevails, jml
On Jul 2, 2010, at 9:08 AM, Jonathan Lange wrote:
My intent is to complete the final steps of the release this weekend. It would be very, very helpful to have someone around who has root on cube and is able to be responsive to requests during a UK afternoon / evening.
I will make an attempt to be available, and on IRC, from 3PM to 9PM United Kingdom Time on the 3rd and 4th of July.
On Fri, Jul 2, 2010 at 2:08 PM, Jonathan Lange <jml@mumak.net> wrote: ...
My intent is to complete the final steps of the release this weekend. It would be very, very helpful to have someone around who has root on cube and is able to be responsive to requests during a UK afternoon / evening.
Currently the release is blocked due to an anomaly on the Windows MSI buildslaves. The slaves are building MSIs that are named with "10.0.0" rather than the expected "10.1.0". I don't have anything locally that will analyze the MSIs to see whether this problem goes deeper than the name of the file, but 'strings' hints that it's likely. See http://buildbot.twistedmatrix.com/builders/winxp32-py2.5-msi/builds/162 for more details. jml
On Jul 3, 2010, at 12:10 PM, Jonathan Lange wrote:
On Fri, Jul 2, 2010 at 2:08 PM, Jonathan Lange <jml@mumak.net> wrote: ...
My intent is to complete the final steps of the release this weekend. It would be very, very helpful to have someone around who has root on cube and is able to be responsive to requests during a UK afternoon / evening.
Currently the release is blocked due to an anomaly on the Windows MSI buildslaves.
The slaves are building MSIs that are named with "10.0.0" rather than the expected "10.1.0". I don't have anything locally that will analyze the MSIs to see whether this problem goes deeper than the name of the file, but 'strings' hints that it's likely.
See http://buildbot.twistedmatrix.com/builders/winxp32-py2.5-msi/builds/162 for more details.
It looks like <http://buildbot.twistedmatrix.com/builders/winxp32-py2.5-msi/builds/162/step...> is identifying the version number for Twisted incorrectly. I note that the log doesn't appear to have a PYTHONPATH entry for "C:\twistedbot\winxp32-py2.5-msi\Twisted", which is where the code actually lives; it's apparently just picking up the system Twisted that buildbot is using. Does anybody have access to this machine to fix this build step?
Le samedi 03 juillet 2010 à 13:29 -0400, Glyph Lefkowitz a écrit :
On Jul 3, 2010, at 12:10 PM, Jonathan Lange wrote:
On Fri, Jul 2, 2010 at 2:08 PM, Jonathan Lange <jml@mumak.net> wrote: ...
My intent is to complete the final steps of the release this weekend. It would be very, very helpful to have someone around who has root on cube and is able to be responsive to requests during a UK afternoon / evening.
Currently the release is blocked due to an anomaly on the Windows MSI buildslaves.
The slaves are building MSIs that are named with "10.0.0" rather than the expected "10.1.0". I don't have anything locally that will analyze the MSIs to see whether this problem goes deeper than the name of the file, but 'strings' hints that it's likely.
See http://buildbot.twistedmatrix.com/builders/winxp32-py2.5-msi/builds/162 for more details.
It looks like <http://buildbot.twistedmatrix.com/builders/winxp32-py2.5-msi/builds/162/step...> is identifying the version number for Twisted incorrectly. I note that the log doesn't appear to have a PYTHONPATH entry for "C:\twistedbot\winxp32-py2.5-msi\Twisted", which is where the code actually lives; it's apparently just picking up the system Twisted that buildbot is using.
Does anybody have access to this machine to fix this build step?
The build step is in the master on cube. I think I've fixed it, sorry for the trouble. -- Thomas
On Sat, Jul 3, 2010 at 9:43 PM, Thomas Hervé <therve@free.fr> wrote: ...
The build step is in the master on cube. I think I've fixed it, sorry for the trouble.
Thanks. It looks like they're working now. What was the problem? If something like this happened again, is there anything we can do other than ask for help and then wait? jml
Le dimanche 04 juillet 2010 à 10:05 +0100, Jonathan Lange a écrit :
On Sat, Jul 3, 2010 at 9:43 PM, Thomas Hervé <therve@free.fr> wrote: ...
The build step is in the master on cube. I think I've fixed it, sorry for the trouble.
Thanks. It looks like they're working now.
What was the problem? If something like this happened again, is there anything we can do other than ask for help and then wait?
The problem was in the msi builder class in the buildbot master. The specific files holding those objects is /srv/bb-master/BuildBot/master/twisted_factories.py This is the interesting part: class TwistedBdistMsiFactory(TwistedBaseFactory): treeStableTimer = 5*60 uploadBase = 'public_html/builds/' def __init__(self, source, uncleanWarnings, platform, pyVersion): python = self.python(pyVersion) TwistedBaseFactory.__init__(self, python, source, uncleanWarnings) self.addStep( LearnVersion, python=python, package='twisted', workdir='Twisted') The problem was that the step used to specify "source" as workdir, instead of "Twisted". The "source" directory probably doesn't even exist, so it didn't get the version from the correct place. It got the wrong value because this step was wrongly copied from PyOpenSSLBuildFactoryBase class. Once fixed, a "make restart-master" in /srv/bb-master/BuildBot/ takes the changed file into account. Hope this helps, -- Thomas
participants (3)
-
Glyph Lefkowitz -
Jonathan Lange -
Thomas Hervé