[Biojava-dev] Rationalisation
mark.schreiber at group.novartis.com
mark.schreiber at group.novartis.com
Tue Feb 24 20:19:31 EST 2004
OK -
Maybe it's a hack and slash approach. I think for BJ2 there is a good
argument for splitting Sybolic and Ontology APIs from the Bio stuff.
Generating too many JARs is a bad idea.
Actually I think that splitting BioSQL support is a good idea (and maybe
merging with DAS, Ensembl, etc). Not everyone uses it and it would give
you the option of getting the BioSQL jar that supports the schema version
you want.
- Mark
David Huen <david.huen at ntlworld.com>
02/24/2004 04:59 PM
Please respond to smh1008
To: Mark Schreiber/GP/Novartis at PH
cc:
Subject: Re: [Biojava-dev] Rationalisation
On Tuesday 24 Feb 2004 3:15 am, mark.schreiber at group.novartis.com wrote:
> As you all know biojava is quite big. It currently splits itself into
two
> jar files (biojava.jar and grammars.jar) or three if you include the
> bytecode.jar.
>
> There have been suggestions of splitting it into subprojects. One
obvious
> place for a split would be biojava.jar and grammars.jar. The ant build
> already achieves this artificially by splitting the code base in two.
> Would it be sensible to give grammars it's own cvs home?
>
> Other possible splits:
>
> demos
> doclets
> apps
>
> My reasoning is as follows. Projects should be merged where components
> have a high level of two way dependency and seperated where there is
only
> one way dependency.
>
> For example biojava depends heavily on grammars.jar but grammars.jar
> doesn't care too much about the core of biojava. Demos and apps are
> dependent on biojava but the reverse is not at all true. The doclets
> project is a nice add on but they are pretty indenpendent of each other.
> My feeling is that this would make management of the project much easier
> and ant scripts simpler.
>
> Any thoughts or flames??
>
By that rationale, classes related to bioSQL would sensibly lie outside
the
main package, as would the DP stuff. Perhaps we could put all the SQL
implementation stuff (bioSQL, Ensembl) into one jar and things like DP
into
another exotics-that-will-do-your-mind-in jar.
My only reservation aobut all this is whether it would make for (perhaps)
more support work when users get confused as to which jars they should
have
had in their classpath.
Regards,
David Huen
More information about the biojava-dev
mailing list