Some tests failed when Python built without docstrings (--without-doc-strings options). Proposed patch fixes most of tests.

Only doctests in test_generators and test_genexps don't fixed. I don't know how to make doctests conditional.

[135/372] test_generators
**********************************************************************
File "/home/serhiy/py/cpython-without-doc-strings/Lib/test/test_generators.py", line ?, in test.test_generators.__test__.email
Failed example:
    print(i.__next__.__doc__)
Expected:
    x.__next__() <==> next(x)
Got:
    <BLANKLINE>
**********************************************************************
1 items had failures:
   1 of  31 in test.test_generators.__test__.email
***Test Failed*** 1 failures.
test test_generators failed -- 1 of 287 doctests failed

[137/372/1] test_genexps
**********************************************************************
File "/home/serhiy/py/cpython-without-doc-strings/Lib/test/test_genexps.py", line ?, in test.test_genexps.__test__.doctests
Failed example:
    print(g.__next__.__doc__)
Expected:
    x.__next__() <==> next(x)
Got:
    <BLANKLINE>
**********************************************************************
1 items had failures:
   1 of  75 in test.test_genexps.__test__.doctests
***Test Failed*** 1 failures.
test test_genexps failed -- 1 of 75 doctests failed

I've already committed a decorator in 5c7f92bfe785, but it isn't
quite robust. I think the one in issue17041-decorator.diff should
do the trick.

Also, we then can use support.HAVE_DOCSTRINGS for some doctests.

Oh, I see Stefan already fixed some failures. Here are updated patches.

Once upon a time (two years ago?) we fixed the tests so that they ran successfully (skipped when appropriate) with -OO set, which omits docstrings.  We were checking for the optimization level (sys.flags.optimize) then.  It seems like it would make more sense to combine both checks into one decorator.

Patches updated incorporating Stefan's patch.

> It seems like it would make more sense to combine both checks into one decorator.

These are different cases. @unittest.skipIf(sys.flags.optimize >= 2) is about docstrings in Python implemented modules, and @support.requires_docstrings is about docstrings in C implemented modules.

Serhiy Storchaka <report@bugs.python.org> wrote:
> > It seems like it would make more sense to combine both checks into one decorator.
> 
> These are different cases. @unittest.skipIf(sys.flags.optimize >= 2) is about docstrings in Python implemented modules, and @support.requires_docstrings is about docstrings in C implemented modules.

But they're always used together, aren't they? I kind of like the idea of a
single decorator.  We could have two different messages "compiled without docstrings"
and "docstrings are omitted with -O2 and above", where the first message
should get priority if both conditions are true.

Given the mood on python-dev concerning --without-doc-strings, I wonder if
we should stick to minimal changes and just skip the tests in test_pydoc.py
instead of fixing them.

Personally I'm happy either way.  David, what do you think?

The patches are really small, so +1 for committing this before the
rc1 for 2.7.4 is released.

I left some comments but have no indication that they got mailed.

Stefan: as Serhiy pointed out, the two parameters control different things.  While it may be true that it is *likely* that someone would use both if they use one, it is not guaranteed.  I think it is better to keep them separate, since while they are related operationally they are otherwise orthogonal, so linking them in the test code just seems wrong :)

Right, keeping the cases separate is indeed clearer.

Serhiy, in case you are waiting for my "stick to minimal changes" comment to be
resolved:  People apparently *do* use --without-doc-strings, so the additional
code in test_pydoc.py is really fine.

New changeset 6c0d9d1ce63e by Serhiy Storchaka in branch '2.7':
- Issue #17041: Fix testing when Python is configured with the
http://hg.python.org/cpython/rev/6c0d9d1ce63e

New changeset 241b655c23b6 by Serhiy Storchaka in branch '3.2':
Issue #17041: Fix testing when Python is configured with the
http://hg.python.org/cpython/rev/241b655c23b6

New changeset b2fa459d95d6 by Serhiy Storchaka in branch '3.3':
Issue #17041: Fix testing when Python is configured with the
http://hg.python.org/cpython/rev/b2fa459d95d6

New changeset 754112dc3bbb by Serhiy Storchaka in branch 'default':
Issue #17041: Fix testing when Python is configured with the
http://hg.python.org/cpython/rev/754112dc3bbb

Thank you, Stefan. I got notification about your comments on email and fixed patches.

Here is a patch which fixes doctests (for 2.7 only need replace test.support by test.test_support). I'm not sure this is a right way to do conditional doctests.

doctests_without_docstrings.patch looks good to me.

Serhiy, if you have time, I think it would be nice to get the remaining
fix into the upcoming release candidates.

New changeset df9f8feb7444 by Serhiy Storchaka in branch '2.7':
Issue #17041: Fix doctesting when Python is configured with the
http://hg.python.org/cpython/rev/df9f8feb7444

New changeset 9c0cd608464e by Serhiy Storchaka in branch '3.2':
Issue #17041: Fix doctesting when Python is configured with the
http://hg.python.org/cpython/rev/9c0cd608464e

New changeset 886f48754f7e by Serhiy Storchaka in branch '3.3':
Issue #17041: Fix doctesting when Python is configured with the
http://hg.python.org/cpython/rev/886f48754f7e

New changeset e6cc582cafce by Serhiy Storchaka in branch 'default':
Issue #17041: Fix doctesting when Python is configured with the
http://hg.python.org/cpython/rev/e6cc582cafce

Thank you for review, Stefan Krah.