[Python-Dev] [Python-checkins] Python Regression Test Failures basics (1) (original) (raw)

Benjamin Peterson musiccomposition at gmail.com
Tue Sep 16 00:13:21 CEST 2008

On Mon, Sep 15, 2008 at 7:04 AM, Nick Coghlan <ncoghlan at gmail.com> wrote:

Neal Norwitz wrote:

On Sun, Sep 14, 2008 at 5:24 AM, Benjamin Peterson <musiccomposition at gmail.com> wrote:

On Sun, Sep 14, 2008 at 4:07 AM, Nick Coghlan <ncoghlan at gmail.com>

What environment are you using to run the debug-mode regression tests? The above four tests run without any problems for me, but I'm just running them in a normal Kubuntu desktop shell. Something in Neal's build which has made a difference before is that the tests are run after a "make install". Benjamin is correct. See Misc/build.sh for the script that generates this. See http://docs.python.org/dev/results/ for details about the most recent run. Hmm, even after doing a make altinstall (don't want to mess with the system Python!), I still can't reproduce those failures. I had to give the test suite elevated privileges to get testdistutils and testtcl to work properly, and this machine has some audio issues that upset testlinuxaudiodev and testossaudiodev, but the four tests that are failing in Neal's tests all run fine. Are these being run in a sub-environment of some kind to avoid messing up the machine, and that environment is missing the /dev/* tree from the filesystem? (the particular error being triggered by the testmultiprocessing failure indicates that as a possible cause, and the testioctl problem relates to not being able to find /dev/tty).

I saw the error that Neal is getting once, and that was when I interrupted test_normalization while it was fetching NormalizationTest.txt. So maybe he just has a corrupt copy of that? (Neal, will you try removing it?)

Since these tests are an additional kind of buildbot, I'd really like to see them all passing before rc2 goes out, but I think I've reached the limits of what I can figure out from this side of the Pacific.

-- Cheers, Benjamin Peterson "There's no place like 127.0.0.1."

More information about the Python-Dev mailing list