[Bioperl-l] Bio::FeatureHolderI

[Lincoln Stein]

I agree with Ewan and Jason that internal filters allow you the efficiency of 
implementing the filter in SQL.  I thought of proposing this, but I don't 
want to add something this radical to the API just before 1.2.  Another issue 
is that if we're going to implement internal filters, then we should do it 
right, allowing for filtering by location, feature type, source tag, etc., 
and for arbitrary boolean combinations of these attributes.

Another thought that occurred to me is that it would be very nice to be able 
to pass in a subroutine that is simply passed each feature, without the 
features being stored in a memory.  This sort of functional programming is 
very nice when the features are coming from a SAX-based XML parser, where it 
is difficult, if not impossible, to implement a next_feature() style iterator 
in the caller.

Lincoln

On Monday 18 November 2002 08:37 pm, Ewan Birney wrote:
> On Mon, 18 Nov 2002, Hilmar Lapp wrote:
> > You could clearly do that via -filter. Are you suggesting -location
> > as a parameter as well?
> >
> > The problem with supporting internal filters (implemented by the
> > feature holder) vs external filters (you supply the filtering method
> > or object) is that internal filters can become unnecessarily complex
> > fairly quickly if they are to be flexible.
>
> But the benefit about internal filters is that it allows different
> implemetnations to optimise it differently (in particular in teh SQL
> issueing or something similar)