Bug#641485: RE : Bug#641485:

[Antonio Valentino]

Hi Yaroslav,

Il 23/01/2012 05:41, Yaroslav Halchenko ha scritto:
> committed/uploaded to sid -- apparently I am not a part of
> debian-science team ;-)  will join and push then
> 
> as for testing -- my tired brain skipped the fact that we invoke testing
> via python -c '.. .test()' so failures if occur do not cause process to
> exit with non-0 status anyways atm... unfortunately upstream swallows
> all the exceptions in tables/tests/test_all.py:test and doesn't return
> anything but None :-/ so we would need either to patch or to make it
> friendly to nosetest -- iirc for me it was causing it to run them
> multiple times (I think because it was "testing" test() itself, so we
> would need to assign __test__=False to it or smth like that).
> 
> cheers and thanks!
> 

I'm investigating on this right now.
Probably I will push some improvement upstream pretty soon.


> On Sun, 22 Jan 2012, Yaroslav Halchenko wrote:
>> +   -for v in $(PYVERS); do \
> 
>> for testing -- in numpy pkg '-' was added intentionally  to allow tests to fail
>> but nevertheless have pkg built.  Are you going to eye gaze at all those 10
>> architectures build log files and look for failed tests?  ;-)  I doubt, so I
>> just went ahead and committed a little  tune up (will push later on if
>> everything looks fine) -- now it will fail the build if tests fail.   But
>> that also would mean that we would need to fix pytables in those cases, I hope
>> you are up for that portion of maintenance duties ;-)
> 

Oh yes, you are right, I forgot to remove the "-"
Thanks for the fix.


cheers

-- 
Antonio Valentino