Re: Test suite regressions (=> Jacl?)

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Jeff" == Jeff Sturm <jsturm@one-point.com> writes:

Jeff> Tclblend: Probably not, at least until someone replaces the JNI
Jeff> code with CNI equivalents.  (In theory libgcj supports JNI, but
Jeff> I haven't had great luck with it.)

Our JNI implementation is complete (except for the invocation API),
but not heavily tested.  My experience though is that bugs in the JNI
wrapper code tend to be very easy to fix.  Please report any JNI bugs
you find, preferably to Gnats.

I don't actively use JNI so I'm unlikely to find any JNI bugs.
However, I'm willing to fix the ones that are reported.

For those who use JNI regularly, I'd like to point out that gcj is
probably the easiest way to write JNI.  That's because you can use gdb
to debug your JNI code and (more-or-less) seamlessly pass between your
Java and JNI code.

Tom