Re: HTSeq

[Andreas Tille <andreas@an3as.eu>]

Hi Diane,

thanks for your effort in HTSeq.

On Mon, May 19, 2014 at 04:20:51PM -0700, Diane Trout wrote:
> I finally had some time to work on figuring out how to write a get-orig-source 
> target for HTSeq.
> 
> The subversion repository includes 69 mbytes of example data,

Compresed or uncompressed.  Typically sequence data can be squeezed
reasonably.

> which upstream 
> uses for test cases. But the official pypi release deletes the test code & 
> test data. (In addition to not shipping the source files for the swig and 
> cython components)
> 
> I'm wondering if I should include the test data in the repack so I can run the 
> tests during build, or if that's too much file bloat?

I *personally* totally convinced that if we want to support scientists
seriously we should try all means we have to test the functionality of a
package.  All packages I recently created vom scratch are featuring test
data in some arch=all *-data or *-examples package and are running the
test suite at build time (via dh_auto_test) and have also autopkgtest
enabled.  I consider this a great success.  My current work on
python-cogent will also go into this direction (but it turned out more
complex than in BioPython case where I also recently added as many test
features as possible).

If you want to backup this personal opinion of mine in general Debian it
might make sense to discuss this on debian-devel (or perhaps as an
intermediate step on Debian Science).

Kind regards

       Andreas.

-- 
http://fam-tille.de