[Biojava-l] RE: Re: JDK1.4 and IDE's

[Thomas Down]

I'm afraid that, realistically, using generics in BioJava
might be some way off -- Unless Sun are moving their release
schedule up a couple of gears, it's going to be 18 months
before JDK1.5 hits the streets

Of course we could adopt the generic types much sooner
(now! :) using the experimental compiler (which works
find for me), but this would upset quite a few people,
especially those who are using IDEs.


Sequences validated at compile time using the generic types
are an interesting idea, but not really compatible with
BioJava's Alphabet model (which is fundamentally dynamic).
I think you'd need a language with a much broader interpretation
of what a `type' means before this could be made to work
(but I'd be very happy if this turns out not to be the
case!)

     Thomas.




On Wed, Jun 06, 2001 at 04:25:44PM +0100, Phillip Lord wrote:
> 
>         Using the generic type system is going to be a fascinating
> thing to do, and is probably going to require substantial changes to
> the biojava interface, and some of the implementations I think. The
> key problem at the moment is that the Element alphabets are currently
> represented as objects, rather than as a specific class, so its not
> possible to use them generically as types. Also some of the
> implementations check that an element is valid part of a sequence
> Alphabet at runtime, whilst genericity should allow this to be done at
> compile time. 
> 
>         It would be interesting to know when the planning for this work
> starts (which I known will probably not be for some time). I currently
> have a fair bit of Java which relies on my own sequence API, and I
> have been considering both porting it to biojava, and/or using
> generics within it. To be able to do both at the same time would save
> me a lot of effort. And of course provide me with the impetus to get
> off my back side and actually do the work! Although as I generally
> code sitting down this is probably a bad metaphor.