[Bioperl-l] Re: LocationI implementation details

[Hilmar Lapp]

So long as the API definition stays on the interface, no particular 
reason other than historical ones (as so often with Bioperl): 
originally we didn't have Atomic.pm from which now all location impls 
inherit (they do, right?) and the coordinate policy was introduced to 
the API some time after inception, so the easiest way to add the method 
and provide all classes with an implementation was to decorate the 
interface.

In the (extremely unlikely) event that somebody has written a LocationI 
implementation without inheriting from Atomic.pm and took advantage of 
the interface implementing the coordinate policy, he/she'd be screwed 
if you move it to Atomic.pm. So maybe just announce on the mailing list 
that you'll do this and who needs to pay attention.

	-hilmar


On Sep 8, 2005, at 1:49 PM, Jason Stajich wrote:

> Hilmar -
>
> Is there any reason for Bio::LocationI to have the coordinate policy 
> implementation?  I have moved the implementation to Atomic.pm in my 
> code with no effects on any of the tests.
>
> I don't like having the implementation there because it assumes the 
> object is represented by a Hash and I think that some ways we could 
> achieve speedier objects are to make some Location and feature objects 
> with arrays underneath (theoretically)...
>
> -jason
> --
> Jason Stajich
> Duke University
> http://www.duke.edu/~jes12
>
>
>
-- 
-------------------------------------------------------------
Hilmar Lapp                            email: lapp at gnf.org
GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
-------------------------------------------------------------