[Biojava-dev] Proposal to change DP interface

[mark schreiber]

Hi -

I guess it's unlikely that people have implemented their own trainers so if
nobody shouts out soon you should change it.

- Mark

Ps you wouldn't mind making some tutorials for them would you?

 

-----Original Message-----
From: biojava-dev-bounces at portal.open-bio.org
[mailto:biojava-dev-bounces at portal.open-bio.org] On Behalf Of David Huen
Sent: Tuesday, 9 December 2003 12:14 a.m.
To: mark schreiber; biojava-dev at biojava.org
Subject: Re: [Biojava-dev] Proposal to change DP interface

On Monday 08 Dec 2003 9:27 am, mark schreiber wrote:
> Hi -
>
> Would it be possible to overload the method and keep the old one as well?
> You could throw a bioexception if the old method is used for a single 
> head model if you wish.
>
OK, I'll do it that way.  It's quite easily accomodated.

There is an equivalent issue in AbstractTrainer:-

singleSequenceIteration(ModelTrainer trainer, SymbolList symList)

to:-

singleSequenceIteration(ModelTrainer trainer, SymbolList [] symList)

If I modified AbstractTrainer to be able to cope with either single or
twohead models then modified the classes that derive from it in BJ to the
new API, would I need to keep the old method in here too?  i.e.  has anyone
written their own AbstractTrainer-derived classes? (In one sense the name of
this method is no longer appropriate.  Would something like
SingleCaseIteration be better but that may also break things).

I'm doing this to try to get in a single/twohead Viterbi trainer.

Regards,
david huen



_______________________________________________
biojava-dev mailing list
biojava-dev at biojava.org
http://biojava.org/mailman/listinfo/biojava-dev