[Biocorba-l] Implementation woes

[Jason Stajich]

Consider this more of a buffer that I had open as I was trying to do the
bioperl implementation.

I am trying to do the bioperl implementation.  Probably mostly only
answerable by Ewan, but can at least see the problems I think I am running
in to.

bsane::seqcore::SeqFeature implements Annotatable.

In Bioperl, SeqFeatures can be arranged in a hierarchy through a
sub_SeqFeature method - am I to assume that in SeqFeature context
get_annotations returns an AnnotationCollection which will be a
collections of SeqFeatures?  Otherwise it is just going to return null and
there won't be a way to get the sub_features.

Implementing it as anything other than an iterator of features is going to
be a bit of a pain from what I can tell.

I'm also not clear if the fact that SeqFeatureCollection implements the
AnnotationCollection interface if this means there are annotations about
this seqfeaturecollection (in Bioperl the seqfeature collection IS the
sequence so this can just return the same thing the sequence did) or if
the get_annotations is supposed to return something to do with actual
Sequence Features.  Perhaps I should peruse the diagrams better.


Ewan do you have any other ideas about this implementation?


It doesn't look like our Bio::AnnotationI is going to map very well into
the BSANE Annotation interface - I'm not even sure any get_value is going
to work for us so I'm planning on not returning anything for that.  I've
decided to just build a simple object that takes -name and -basis as
arguments and will return these - we don't handle basis anywhere in
bioperl so I'm not sure how we'll get that in properly.


I am realizing better the desire to not make everything Annotatable...
this is probably not going to get implemented for objects that it doesn't
make sense for- not sure what it means for the sequence database to be
annotateable in bioperl...


Finally - with CORBA::ORBit how can you have it bring in > 1 IDL - or do
we just create a single file of all of the IDLs together?  This seemed to
work.

-jason
-- 
Jason Stajich
Duke University
jason@cgt.mc.duke.edu