Re: gij gcj-dbtool and questions

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Jakob" == Jakob Praher <jpraher@yahoo.de> writes:

Jakob> Yeah I thought there is some magic involved. Allthough at first I hoped
Jakob> there were a libtree somewhere :-) so that you could register the hooks
Jakob> dynamically. I also noticed that there are no Makefiles (which is clear,
Jakob> since the tree should be part of the gcc/subdir/Make-lang.in.

Yeah.  Inspired by your note, I went ahead and made a "gcjx-branch"
and put everything there.  You still can't build the tree part of
gcjx, though, since I haven't finished the build hacking.  You can
build gcjx itself though.

This isn't fully useful since I recently discovered a serious bug in
the bytecode back end.  I'm not sure how I'm going to go about
communicating gcjx status... a web page?  Just some file in CVS?

Jakob> But I have to say that the structure of the code is pretty good.

Thank you.

Jakob> What's the status of getting c++ frontends into gcc?

gcjx is the guinea pig for this :-).  I think to make this really work
we need the "top level bootstrap" thing that folks have talked about
for a while.  The idea, as I understand it, is to build g++ and
libstdc++, then go back and build gcc again, this time including gcjx.

Jakob> This partially implemention of C++ member functions is quite handy for
Jakob> this kind of scenario, though a bit hard to find which translation unit
Jakob> implements what.

Yeah.  There is a mini tour of the source in gcjx/PROJECTS, near the
end.  Let me know if I can make it simpler to understand.

Tom