> Here is a short list of what I can think of now. Deferring class
parsing
> is a good idea. it should boost the speed.

Definitely. classes()/superclass()/etc should all consult JavaParser if
not scanned yet, otherwise return the scanned class. Such tricks are
also possible for tags, parameters and so on. A user rarely modifies
many classes so excluding classes from javadoc loading/parsing can make
it super fast. I'm sure we can make it n times faster where n>10 :o)

> 2) if a class implements an interface only available as binary, it
seems
> it's not discovered. see xdocletgui. nothing shows up under tag
libraries.

I'll investigate it.

> Enhancements that can be done.
> 1) simplify the java grammar. it's useless/time consuming to parse the
> method bodies.

Can we ignore method bodies? I think we can only prevent jjtree from
creating various AST classes/instances such as
blablaExpression/SynchronizedBlock/etc. Right?

> 2) general javacc optimisations, like making parser static etc.

Yup, JavaParser should be a utility class, probably with a repository of
already scanned classes (assuming lazy scanning classes is used).

> structural enhancements/refactorings
> 1) expose less of the api. make more classes/methods package private.
> 2) remove the classicwrapper/xml packages. they were for testing at an
> early stage, but the junit stuff is sufficient now.

Ara.


_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com


_______________________________________________
Xdoclet-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/xdoclet-devel

Reply via email to