Re: Patch: g++ -vs- Java exceptions, take 2

[Tom Tromey <tromey at cygnus dot com>]

>>>>> "Gerald" == Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:

Gerald> Sorry, I didn't mean to refer to the patch, but I thought that
Gerald> you suggested a new 2.95.x release containing the current Java
Gerald> frontend.  Making such an extensive change is something we
Gerald> haven't made on release branches in the past.

Yeah, I understand.

Gerald> I definitely see the problem. I'm just not sure what the best
Gerald> solution is.  If you consider the Java stuff in GCC 2.95.2
Gerald> very broken and that upgrade does not effect anything else...

It isn't that what is there is broken per se.  The problem is that the
runtime library has made great advances since the 2.95.2 release --
and the new library requires the new compiler in order to work.  The
new library is so different that there's no real comparison (e.g., the
current runtime has a Java bytecode interpreter now) and in practice
basically everybody wants to use it.

So yes, we've been hoping that since massive changes to gcc/java won't
affect non-Java code, that we can put the current Java front end into
2.95.3.  One of the Java contributors, Bryce McKinlay, has already
done the work of porting the new code into the 2.95.2 tree; he
regularly advises people to go this route to avoid problems with the
cvs tree.

Tom