Opened at 2009-05-21T19:36:36Z
#21 new defect
"setup.py flakes" emits spurious output, confuses buildbot step
Reported by: | warner | Owned by: | |
---|---|---|---|
Priority: | minor | Milestone: | |
Version: | 0.5.1 | Keywords: | pyflakes |
Cc: | Launchpad Bug: |
Description
I think that "python setup.py flakes" should behave the same way as running regular "pyflakes" would: if there are no problems with the source code, it should emit no output.
When all is well, "python setup.py flakes" emits a "running flakes" message, even when --quiet is added. But most of the time, I see it emit additional noise, some sort of setuptools stuff:
Installed /Users/bb-pycryptopp/stuff/slave-pycryptopp/mac-i386-osx-10.5-faust/build/setuptools_darcs-1.2.8-py2.5.egg Searching for setuptools-pyflakes>=1.0.0 Reading http://pypi.python.org/simple/setuptools_pyflakes/ Reading http://allmydata.org/trac/setuptools_pyflakes Best match: setuptools-pyflakes 1.0.5 Downloading http://pypi.python.org/packages/source/s/setuptools_pyflakes/setuptools_pyflakes-1.0.5.tar.gz#md5=09b6ccdb0cb05c5120e27c155ed54914 Processing setuptools_pyflakes-1.0.5.tar.gz Running setuptools_pyflakes-1.0.5/setup.py -q bdist_egg --dist-dir /tmp/easy_install-_F4z9e/setuptools_pyflakes-1.0.5/egg-dist-tmp-KSOwRj Installed /Users/bb-pycryptopp/stuff/slave-pycryptopp/mac-i386-osx-10.5-faust/build/setuptools_pyflakes-1.0.5-py2.5.egg Searching for darcsver>=1.2.0 Reading http://pypi.python.org/simple/darcsver/ Reading http://allmydata.org/trac/darcsver Best match: darcsver 1.2.1 Downloading http://pypi.python.org/packages/source/d/darcsver/darcsver-1.2.1.tar.gz#md5=88e400cd49bb2b72947af7b40601d256 Processing darcsver-1.2.1.tar.gz Running darcsver-1.2.1/setup.py -q bdist_egg --dist-dir /tmp/easy_install-TQMvPQ/darcsver-1.2.1/egg-dist-tmp-J4MICP Installed /Users/bb-pycryptopp/stuff/slave-pycryptopp/mac-i386-osx-10.5-faust/build/darcsver-1.2.1-py2.5.egg running flakes
(this is on the new "faust" buildslave, but I see it appear on other buildslaves too).
This may be related to the setuptools problem in which certain platforms rebuild dependent libraries even though they've already been built once, because the second pass doesn't look in the correct spot.
The consequence is that the buildbot "pyflakes" step never reports success: the spurious output lines look like pyflakes warnings.