jogl package

[Anthony Green]

Here's my first stab at an RPM with the new scripts (find-and-nativify
and rebuild-gcj-db).  It's the OpenGL java bindings:

http://people.redhat.com/green/FC4

A couple of observations....

It's still not super-trivial to add native versions of jar files.  See
attached spec file.  On the other hand, I don't know how we could make
it simpler.

Our libjawt.so.6 is a bit of a problem.  I was unable to run any jogl
code with a non-gcj alternative, since they all pick
up /usr/lib/libjawt.so.6, which has references to our gtk peers and
fails to load.  Should libjawt.so be an alternative?  Or can we install
it somewhere gcj specific?

AG

[Thomas Fitzsimmons]

On Sun, 2005-04-03 at 16:41 -0700, Anthony Green wrote:
> Here's my first stab at an RPM with the new scripts (find-and-nativify
> and rebuild-gcj-db).  It's the OpenGL java bindings:
>
> http://people.redhat.com/green/FC4
>
> A couple of observations....
>
> It's still not super-trivial to add native versions of jar files.  See
> attached spec file.  On the other hand, I don't know how we could make
> it simpler.
>
> Our libjawt.so.6 is a bit of a problem.  I was unable to run any jogl
> code with a non-gcj alternative, since they all pick
> up /usr/lib/libjawt.so.6, which has references to our gtk peers and
> fails to load.  Should libjawt.so be an alternative?  Or can we install
> it somewhere gcj specific?

I think we should rename it libgcj-jawt.so.  libjawt-using apps will
look for $JAVA_HOME/jre/lib/i386/libjawt.so which we already have a
symlink for in java-gcj-compat.

Tom

[Anthony Green]

On Mon, 2005-04-04 at 01:24 -0400, Thomas Fitzsimmons wrote:
> I think we should rename it libgcj-jawt.so.  libjawt-using apps will
> look for $JAVA_HOME/jre/lib/i386/libjawt.so which we already have a
> symlink for in java-gcj-compat.

This is a fine idea for java-gcj-compat users, but what about the
others?

AG

[Anthony Green]

On Mon, 2005-04-04 at 01:24 -0400, Thomas Fitzsimmons wrote:
> I think we should rename it libgcj-jawt.so.  libjawt-using apps will
> look for $JAVA_HOME/jre/lib/i386/libjawt.so which we already have a
> symlink for in java-gcj-compat.

Tom - this doesn't quite work as implemented.

It turns out that jogl does a System.loadLibrary("jawt") which fails
since we renamed the library.

I'm not sure what the best solution is.

If we make libjawt.so an alternative, then it will mess up running non-
alternative java programs.  Maybe we should have stowed it away in a
libgcj specific directory instead of renaming it.

AG

[Thomas Fitzsimmons]

On Tue, 2005-05-24 at 11:40 -0700, Anthony Green wrote:
> On Mon, 2005-04-04 at 01:24 -0400, Thomas Fitzsimmons wrote:
> > I think we should rename it libgcj-jawt.so.  libjawt-using apps will
> > look for $JAVA_HOME/jre/lib/i386/libjawt.so which we already have a
> > symlink for in java-gcj-compat.
>
> Tom - this doesn't quite work as implemented.
>
> It turns out that jogl does a System.loadLibrary("jawt") which fails
> since we renamed the library.

This should work, if LD_LIBRARY_PATH points
to /usr/lib/jvm/jre/lib/%{_arch}.

Tom

[Anthony Green]

On Tue, 2005-05-24 at 14:56 -0400, Thomas Fitzsimmons wrote:
> This should work, if LD_LIBRARY_PATH points
> to /usr/lib/jvm/jre/lib/%{_arch}.

Ok - that works - but I don't think we should force people to set
LD_LIBRARY_PATH.

Maybe Runtime.loadLibrary could be taught to look there.

AG

[Anthony Green]

On Tue, 2005-05-24 at 12:25 -0700, Anthony Green wrote:
> Maybe Runtime.loadLibrary could be taught to look there.

I've filed a bug for this:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21741

AG