Hi,
> I think I've spotted a bug in the code, though. For each
> method it finds
> whose name matches, it does
> >Parameter[] params = methods[i].parameters();
> >for( int j = 0; j < params.length; j++ )
> >{
> > if( parameters == null || !params[j].typeName().equals(
> parameters[j]
> >) )
> > {
> > continue methodLoop;
> > }
> >}
> Suppose we're looking for foobar(String) and it finds foobar() first.
> params.length will be zero, so won't it skip straight past
> and so generate
> the block for that method instead of the one we're actually
> looking for?
You are right.
>
> > > In which case, how about "forMethod"?
> > > (it's
> > > similar to forAllMethods, but does it for only the one that's
> > > specified rather than for each of them...)
> >
> >... and have 2 forMethod loops, one for the getters and one for the
> >setters ? I think it is better that both are generated
> inside a unique
> >loop : the "entity bean attributes loop".
>
> No need for two loops, since (at least the last time I looked)
> forAllPersistentFields goes through both setters and getters,
> storing each
> field as it goes and skipping if it comes up again. So, if
> there's only a
> setter it should still include it and we can use
>
> forAllPersistentFields
> ifHasMethod(getter)
> forMethod(getter)
> ifIsNotAbstract
> <getter stuff>
> ifHasMethod(setter)
> forMethod(setter)
> ifIsNotAbstract
> <setter stuff>
>
Again right.
> > > Also, what happens if the bean class only defines a setter for a
> > > field and not a getter? Not very likely, I know, but
> possible e.g.
> > > for a password
> > > field. Perhaps the getterMethod bits should also have an
> ifHasMethod
> > > around
> > > them too?
> >
> >Can you do that in cmp 2.0 abstract classes ? If yes then I guess
> >xdoclet need it too, don't see how right now...
>
> Without going an having a look through the spec, I've no
> idea. However, I'm
> still using 1.1 and BMP, and I don't know any reason it can't
> be done with
> those. So long as we still allow ejbspec="1.1", I guess we
> ought to cater
> for it.
OK.
>
>
> Andrew.
>
vincent
_______________________________________________
Xdoclet-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/xdoclet-devel
