[Zope-dev] KGS buildbot news

[Marius Gedminas]

News about the KGS buildbot at http://zope3.pov.lt/buildbot/:

  * all builds are done in a virtualenv sandbox uncontaminated by system
    python packages
  * the waterfall display shows a brief summary (ntests/errors/failures)
  * there are short logs with summaries only (lists of tests that
    failed, had errors, or could not be imported)
  * there are bookmarkable links to the latest summaries (and logs) at
    http://zope3.pov.lt/    

Latest test results

  * Python 2.4, 32-bit Linux:

    - zope.server tests fail horribly with invalid file descriptor
      errors.  I cannot reproduce this error in isolation (bin/test -m
      zope.server), but I can reproduce it when I run all the tests
      together.  Some previous test is mucking stuff up.

    - z3c.rml depends on PIL but doesn't express that in its setup.py.
      It also depends on Reportlab.  Neither PIL not Reportlab can be
      installed as eggs from PyPI.  There are working eggs for both PIL
      and Reportlab available on the Internet, AFAIK, but so far nobody
      worked with the respective upstreams to get those into PyPI.
      Since I don't use z3c.rml myself, and since repeated import errors
      for every test layer clutter the output a lot, I've disabled
      z3c.rml tests for now.

  * Pytohn 2.4, 64-bit Linux:

    - same as Python 2.4 on 32-bit Linux, plus
    - one or two zdaemon tests fail nondeterministically (the "daemon"
      process, usually sleep 10 or sleep 100, exits before the test can
      invoke "zdaemon stop"  I can reproduce it (nondeterministically)
      on various machines.  I do not understand how "sleep 100" can
      terminate in under 18 seconds that it takes for the tests to fail.
    - zope.app.apidoc, zope.app.renderer and zope.app.generations errors that
      end up as an ImportError: No module named roman inside a docutils.
      I suspect a problem in the environment and I'm tempted to rm -rf
      the buildout working dir and see if I can reproduce it starting
      from a clean state.

  * Python 2.5, 32-bit Linux:

    - same as Python 2.4 on 32-bit Linux, plus
    - z3c.form expects a certain behaviour from Exception.__repr__,
      which changed in Python 2.5

  * Python 2.5, 64-bit Linux

    - a horrifying amount of errors (80 total), I'll retry after
      cleaning up the buildbot slave dir.

Marius Gedminas
-- 
C is a language that combines all the elegance and power of assembly language
with all the readability and maintainability of assembly language.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: Digital signature
Url : http://mail.zope.org/pipermail/zope-dev/attachments/20080808/28187f83/attachment.bin