[Biopython-dev] Biopython 1.49 beta released

[Peter]

On Fri, Nov 14, 2008 at 3:39 PM, Bruce Southey <bsouthey at gmail.com> wrote:
> Peter wrote:
>> We've had a few Numeric -> NumPy bugs reported,
>>
>> http://bugzilla.open-bio.org/show_bug.cgi?id=2658
>> Bug 2658 - Bio.PDB.Neighborsearch
>>
>> http://bugzilla.open-bio.org/show_bug.cgi?id=2649
>> Bug 2649 - Bio.KDTree (probably fixed)
>>
>> ...
>
> I noticed that Bio.PDB.Neighborsearch is not being tested.
>

That fact that we didn't spot Bug 2658 from the unit tests makes that
very clear ;)

>
> Is there someway to identify which functions are not getting tested?
>

I can't think of an easy way - the best bet might be a quick script to
scan all the unit tests and pull out import lines, and from this build
a list of all modules which have some coverage.  This wouldn't tell us
about how much of each module is tested, but it would be better than
nothing.

> I know it is considerable effort but it would allow the development of tests
> that at the very least exercise all the Biopython code. (Hopefully this is
> not as bad as the Numpy documentation marathon.)

I've written plenty of tests myself, including for existing modules -
my gut feeling is full test coverage would be quite a marathon.

Compared to the early years of the project, I've propably tried to be
a bit stricter about making sure we have test cases and documentation
before accepting new code.  In some cases this has worked out pretty
well (e.g. Tiago's PopGen stuff is covered in the tutorial and has
unit tests).  In other cases it could put people off contributing
code.

Peter