[Biojava-dev] (no subject)

[Thomas Down]

On Mon, Oct 21, 2002 at 10:12:51AM +0100, Keith James wrote:
> 
>     Thomas> I was actually using these on Friday, and thought at the
>     Thomas> time that they probably ought to be moved into the main
>     Thomas> tree, in the absense of any easier way to get the
>     Thomas> blast/ssbind stuff up and running.  Would anyone have any
>     Thomas> objections to moving them across?
> 
> Move away, I say...

Okay, these have moved now -- tests still pass, and very little
has changed.  But I no longer have ant-build/classes/tests on my
CLASSPATH when running blast-parsing scripts...

> I was never quite sure what the intention was behind having to use
> SequenceDBs, so I left things as I found them, although it's a right
> pain in the ****. Is this a thing for Biojava 2 (given that it
> involves interface changes) or more immediate, do you think?

Well, we certainly ought to do something different for BioJava2 --
discussions so far have talked about using JNDI naming throughout.
That could probably help in this case -- users will be able to get
at a SequenceDB through JNDI where it's available, without having
to create dummy objects explicitly when it's not.

For 1.3...  Interface-wise, moving from returning SequenceDBs and
Sequences to String-y IDs feels a bit of a step backwards, but
very few people /actually/ seem to be using real SequenceDBs at
the moment, so I guess the current interfaces are just being
a barrier to entry.

Heigh ho...

     Thomas.