[Biojava-l] Re: HMM's and Distributions

Thomas Down td2@sanger.ac.uk
Wed, 17 Jan 2001 10:51:59 +0000


On Wed, Jan 17, 2001 at 02:54:40PM +1300, Mark Schreiber wrote:
> 
> It does raise a point though. At what point will the biojava API become
> stable. ie shouldn't we deprecate outdated methods from now on rather than
> removing them entirely??  It would seem to be breaking one of the
> fundamentals of OO design.

Part of the problem here is that different parts of BioJava have
distinct lifecycles.  There's `core' stuff which is pretty stable --
certainly things like the Alphabet, SymbolList, and Sequence
interfaces really /ought/ to be stable, and we should probably
go the deprecation route if we ever do need to remove old methods.
Hopefully sequence input is reaching this point now, after the
refactoring of the last few months.

The dynamic programming toolkit is still in rather more of a
state of flux, and has changed architecture quite a bit since
it was originally written.  It certainly /will/ stabilize, but I'd
guess it's up to Matthew (who's done almost all the architecture
work in this package) to say when.

That said, I hope there won't be too many more upheavals -- the
current API (while very different from what was released in 1.0)
has been fairly stable for a while now.

Matthew?