Test failures in graph subpackage

[Josh Warner]

Should we temporarily modify travis.yml to install an older version of 
Cython? This bug is getting in the way of the usual Travis workflow, and 
now that we know the source of the problem it seems counterproductive to 
have Travis screaming at everyone. Once we either fix the problem (or 
determine it's a Cython bug), we can revert the travis.yml change.

I'll make a PR for this.

On Monday, April 22, 2013 10:03:24 AM UTC-5, Josh Warner wrote:

I think I found the problem.
>
> Originally, these tests all passed on OSX, Win7, and Scientific Linux. 
>  Then I looked into what versions Travis was installing, and the only 
> package newer than what I was running was Cython 0.19.
>
> So upgraded Cython to 0.19 using pip (installed with no errors; had 0.17 
> or 0.18 previously) with no other changes in my config. Then after `$ make 
> clean`, rebuilt scikit-image inplace, and `$ make install` I can reproduce 
> these failures on Scientific Linux.
>
> I renamed and kept the old _mcp.c file to compare with the new version; if 
> that diff is potentially of interest I can post it or the full file(s) on 
> pastebin or a gist.
>
> Josh
>
> On Saturday, April 20, 2013 3:32:16 PM UTC-5, Tony S Yu wrote:
>>
>> Can anyone reproduce the test failures that the Travis buildbot is 
>> raising on master:
>>
>> https://travis-ci.org/scikit-image/scikit-image/jobs/6472929#L2274
>>
>> The tests pass on my setup, so I can't really figure out what's wrong.
>>
>> Thanks,
>> -Tony
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/scikit-image/attachments/20130422/0f0aba8e/attachment.html>