> From: Petter Reinholdtsen [mailto:[EMAIL PROTECTED]]
>
>
> [John Keiser]
> > In jnimeth.c, it looks like line 73 should say result.unionselector,
>
> Hm, not in my source. There the result is jtype, and gives correct
> return type.
>
Oops, sorry, I read wrong on that one. You're right.
>
> > Also, if you're interested, in libnative/java.lang.reflect, there is
> > a serious error in invokeNative(). It does not handle void methods
> > or methods that return primitive types correctly, instead simply
> > using CallObjectMethod in all cases. I assume you can see how
> > easily this will cause segfaults. This will be fixed in the
> > Japhar+Classpath stuff, though, so unless you are desperate to get
> > java.lang.reflect for Sun working right now, you may as well wait
> > and copy stuff from my patch when it's done.
>
> I'll have a look at this.
>
> BTW: When is the estimated time of arrival for Classpath? I try to
> keep up-to-date with the CVS source, but has been unable to make a
> working .class package.
I estimate about a month for something stable. I'll probably have a patch
for Japhar in a week or two--I'll provide that as soon as I get Hello World
working, and then you and Chris and I can work out how to mesh it in with
the current tree. Brian has Classpath compiling a .jar file, but we need
about 5 more substantial classes and the exception classes before we're
completely standalone. I suspect a working build process will be ready
fairly soon (like a week or less), but Brian is the one to ask on that. I
have a personal copy of classpath.jar that I have been using, but I compiled
that over long hours, not through a Makefile.
--John Keiser
