[Bioperl-l] Devel::Cover

Chris Fields cjfields at uiuc.edu
Thu Mar 8 10:08:12 EST 2007


On Mar 8, 2007, at 8:20 AM, Sendu Bala wrote:

> Nathan Haigh wrote:
>> Find the Devel::Cover report for bioperl-live here:
>> http://www.bioinf.shef.ac.uk/public/bioperl-live/cover_db/ 
>> coverage.html
>
> Thanks for that.
>
>
>> First things to note are:
>> 1) There appear to be quite a few modules without POD for all methods
>> (pod column).
>> 2) The test suite doesn't test all subs in the modules (sub column).
>> More rigorous code coverage would be achieved by ensuring tests were
>> designed to execute all statements (stmt column). Details of the  
>> other
>> columns can be found in Devel::Cover::Tutorial but essentially inform
>> how well all the different possible routes in condition statements  
>> are
>> covered by the tests.
>> Have fun!
>
> Really interesting, but fun? I'm actually kind of frightened by it ;)
> In some crazy dream I'd like to see the pod and sub columns at 100%  
> for all modules in time for Bioperl 1.6. From a brief scan though  
> it seems like an incredible amount of work would be needed.

I'm looking at these not as absolute values as much as a relative  
indicator of areas that need improvement, but the two areas you  
mention (pod, sub) should be higher.  Things like branch and  
condition coverage will be much harder to get to 100% and may be  
unrealistic.

Saying that, I think we could aim for threshold values, like have a  
minimum of ~75-80% for the total, or have a minimal coverage of  
80-90% for specific things like subs, POD, etc, with the gold  
standard being 100%.

> Doing the POD would be relatively easy. Does anyone feel inspired  
> to take on that particular challenge I wonder?

Docs?  That should be up to the developer, but we know how that  
goes  ; >

I think the issue with POD is laid out in the project priority list,  
but maybe it should be moved up.

> PS. Did you run this with BIOPERLDEBUG=1 ? I have t/RemoteBlast.t  
> hanging on me atm.

chris



More information about the Bioperl-l mailing list