[Bioperl-l] Bio::RangeI::union

[Hilmar Lapp]

Sounds good to me.

Interfaces may have decorating methods but we wanted to deprecate that 
interfaces are instantiated. So if RangeI tries to instantiate itself 
then that's not good; no decorating method on an interface should ever 
attempt to instantiate the interface.

	-hilmar

On Jun 10, 2005, at 3:39 AM, Aaron J. Mackey wrote:

> Or, how about deprecating the Bio::RangeI->union() construct (since 
> although we've supplied implementations in interface, we don't need to 
> encourage people to use them)?  This is just a weird way to do 
> $newrange = Bio::Range->union(@ranges), right? (which saves a whole 
> capitalized keystroke!)
>
> -Aaron
>
> On Jun 9, 2005, at 2:45 PM, Chris Mungall wrote:
>
>>
>> The pod docs for union() state that this is is valid:
>>
>>   my $newrange = Bio::RangeI->union(@ranges);
>>
>> In the subroutine body, this gets called:
>>
>>   my $self = shift;
>>   ...
>>   $self->new(...)
>>
>> Since $self is equal to the string "Bio::RangeI", rather than an 
>> object
>> implementing this interface, this will result in a call to
>>
>>   Bio::Root::RootI->new("Bio::RangeI",...)
>>
>> This works fine in bp1.4, but in recent bioperl revisions this 
>> results in
>> a warning message that Bio::Root::RootI->new is deprecated, and a
>> delegation to Bio::Root::Root, **omitting the name of the class to be
>> created**, thus creating a Bio::Root::Root object, which is useless 
>> and
>> will inevitably break any code calling the union() method.
>>
>> I think this delegation is completely wrong, and should be removed, 
>> and
>> the warning message switched to an error; OR it should be 
>> undeprecated and
>> the original behaviour behaviour restored
>>
>> If we decide that RootI->new is truly deprecated, then Bio::RangeI 
>> should
>> have to do some $self examination, and use the correct object
>> instantiation method, rather than $self->new. I don't really know 
>> what the
>> correct object instantiation method is - perhaps just 
>> Bio::Range->new()?
>> Or should a factory be used?
>>
>> Personally, I would prefer it if Bio::RootI->new were undeprecated 
>> and the
>> original behaviour restored. deprecating would make perfect sense if
>> bioperl interfaces really were interfaces, which they are not.
>>
>> Cheers
>> Chris
>>
>> _______________________________________________
>> Bioperl-l mailing list
>> Bioperl-l at portal.open-bio.org
>> http://portal.open-bio.org/mailman/listinfo/bioperl-l
>>
>
> --
> Aaron J. Mackey, Ph.D.
> Project Manager, ApiDB Bioinformatics Resource Center
> Penn Genomics Institute, University of Pennsylvania
> email:  amackey at pcbi.upenn.edu
> office: 215-898-1205
> fax:    215-746-6697
> postal: Penn Genomics Institute
>         Goddard Labs 212
>         415 S. University Avenue
>         Philadelphia, PA  19104-6017
>
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at portal.open-bio.org
> http://portal.open-bio.org/mailman/listinfo/bioperl-l
>
>
-- 
-------------------------------------------------------------
Hilmar Lapp                            email: lapp at gnf.org
GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
-------------------------------------------------------------