[Bioperl-l] Bio::LocatableSeq warning

[Chris Fields]

On Dec 17, 2008, at 6:52 PM, Mark A. Jensen wrote:

>
> ----- Original Message ----- From: "Sendu Bala" <bix at sendu.me.uk>
> To: "Chris Fields" <cjfields at illinois.edu>
> Cc: "bioperl-l" <bioperl-l at lists.open-bio.org>; "Mark A. Jensen" <maj at fortinbras.us 
> >
> Sent: Wednesday, December 17, 2008 6:18 PM
> Subject: Re: [Bioperl-l] Bio::LocatableSeq warning
>
>
>> Chris Fields wrote:
>>> It is a problem with clustalw parsing not accounting for  
>>> frameshifts, mapping, and other odd bits coming from this format.   
>>> That could be added in; LocatableSeq now allows this by specifying  
>>> position => shift in frameshifts() as a hash ref.
>>>
>>> There are other significant problems with LocatableSeq, though.    
>>> Gaps, frameshifts, and residues are set and checked via globals  
>>> and aren't set per instance (which is definitely not optimal).   
>>> I'll file a bug report on this to track it.
>>
>> In this instance though, is there anything actually wrong with the  
>> alignment object I get?
>>
>> If not, perhaps it would be appropriate to revert to the old  
>> behaviour and have the message be a debug() and not a warn()? I  
>> don't know - I didn't consider the implications - just a suggestion.
>>
>
> A softer warning would be convenient. I will frequently set end() just
> to get the side-effect of a correct recalculation of the end coord.  
> (This is kludgy
> in itself, of course.)

It's never a good idea to rely on this behavior, particularly when  
considering the issues with global variables I point out here:

http://bugzilla.open-bio.org/show_bug.cgi?id=2715

We can certainly change the warning to not reset passed end() values,  
but it should definitely remain a warning to indicate a problem that  
needs to be addressed.

chris