[Biojava-dev] Feature filtering changes?

[Thomas Down]

On Fri, Oct 11, 2002 at 12:05:30PM +0100, Matthew Pocock wrote:
> Sounds good to me. The recursive vs flat filtering is
> a bit of a mess right now, and this is at least not a
> step backwards. I like the SequenceDB.filter() method,
> but wish we could be doing queries like this propperly
> over any objects. Ah well. Are you able to write the
> JUnit tests & roll over the biojava objects to the new
> API, or would you like some help?

It's not a big job.  The SequenceDB implementations floating
around are:

   - HashSequenceDB, IndexedSequenceDB, RagBag: can't do much for these,
     they'll just inherit a naive `iterate over all the sequences'
     implementation from AbstractSequenceDB.

   - BioSQLSequenceDB: already done, only change is that the
     existing method will now be callable by interface.

   - DASSequenceDB: The DAS protocol is a bit dumb, so the
     possibilities are limited.  However, there's a `fetch a
     single feature by ID' request in the protocol which is
     useful to some people, so we can expose that efficiently
     via the new API.

Any that I've missed?  None of this should take very long...

I'll also be updating biojava-ensembl to take advantage of
this new call.  I've made a start at this, but there's still
more work to do there.

Unless there are objections, I'll probably check in changes + tests
over the weekend.

    Thomas.