Jp Calderone <exarkun@divmod.com> writes:

Attached is the output from a trial run using Python 2.4c1. There seems to be a new problem in trial's signal handling code.

Got it. Some of the signal-handling code was trying to setup a SIGKILL handler, which is an error at the C level but wasn't reported at the python level until 2.4. The tests are running again, 3 failures and 9 errors, all in test_trial. Sample attached. The buildslave is using 2.4c1 all the time now, right? cheers, -Brian =============================================================================== [ERROR]: testBrokenTearDown (twisted.trial.test.test_trial.FunctionalTest) File "/home/buildbot/run/full2.4/Twisted/twisted/internet/defer.py", line 117, in maybeDeferred result = f(*args, **kw) File "/home/buildbot/run/full2.4/Twisted/twisted/trial/test/test_trial.py", line 126, in testBrokenTearDown imi = itrial.IMethodInfo(self.reporter.udeMethod) exceptions.AttributeError: 'BogusReporter' object has no attribute 'udeMethod' File "/home/buildbot/run/full2.4/Twisted/twisted/internet/defer.py", line 117, in maybeDeferred result = f(*args, **kw) File "/home/buildbot/run/full2.4/Twisted/twisted/trial/test/common.py", line 170, in tearDown self.reporter.verify(self.failIfImportErrors, self.checkReporterSetup) File "/home/buildbot/run/full2.4/Twisted/twisted/trial/test/common.py", line 113, in verify assert_(getattr(self, v), 'self.%s did not evaluate to non-zero' % (v,)) File "/home/buildbot/run/full2.4/Twisted/twisted/trial/assertions.py", line 45, in failUnless raise FailTest, msg twisted.trial.assertions.FailTest: self.setUpReporterCalled did not evaluate to non-zero