[Bioperl-l] bioperl-run parameter question

Chris Fields cjfields at uiuc.edu
Mon Dec 11 20:07:16 UTC 2006


On Dec 11, 2006, at 12:21 PM, Sendu Bala wrote:

> Chris Fields wrote:
>> I am writing up a few bioperl-run modules and have a simple  
>> question,  though I don't know if anyone knows the answer.  I was  
>> curious as to  why parameters for most (all?) bioperl-run modules  
>> lack the '-'  preceding them.  This came up re: StandAloneBlast  
>> last week  (something Torsten fixed), but I noticed just about  
>> every bioperl-run  module uses the dashless parameters.
>
> I didn't follow that particular thread, but from my experience  
> there is a useful distinction between bioperl options using the -  
> as normal for full consistency with core (eg. -verbose), whilst the  
> options that belong to the program the run module is a wrapper for  
> do not take dashes. Again, this seems consistent within the run  
> package.

I respectfully disagree that this is a 'useful' distinction.  My main  
point is consistency.  To me, it's counterintuitive to have two  
Bioperl classes, both which inherit Bio::Root::Root, use two  
different syntaxes for any parameters passed to the constructor, even  
if some are 'program' parameters.  It's also not consistent with  
StandAloneBlast or RemoteBlast, both which are considered bioperl-run  
modules even though they are in core, and both or which use dashed  
parameters (StandAloneBlast actually allows both).  In fact, it isn't  
consistent within bioperl-run itself.   
Bio::Tools::Run::EMBOSSApplication uses dashes for parameters in a  
hashref!

Okay, judging by the previous examples, 'consistency' isn't a word I  
would use to describe bioperl-run as a whole (back to Jason's 'cat- 
herding' analogy).  It would be easier to let it slide for now,  
especially since changing them would be a serious pain, not to  
mention an API issue.  But shouldn't there be some consistency?

And what about new modules?  Do we follow the historical (possibly  
confusing) 'dashless' route, or use the core-like dashed approach  
(thus breaking from the other run modules)?

> I'd suggest sticking to the current pattern.
>
>
> Cheers,
> Sendu.

I'll allow for both, ala StandAloneBlast.  Doesn't hurt to be safe. ; >

Have fun at the hackathon!

chris



More information about the Bioperl-l mailing list