GCJ/minGW produced executables and linux/wine

[Tom Tromey]

>>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:

Andrew> [ smart pointers ]
Andrew> IMO we should do this, but everyone else seems worried that it
Andrew> involves changing CNI.  That doesn't seem such a bug deal IMO.  It
Andrew> would make gcj run correctly on far more targets.

My real problem with it is just that it requires a flag day for the
whole library.  Plus, I just don't want to change all that code -- how
horrible.

Still, given that the g++ patch was rejected, I don't think we have
another sensible choice.  It just requires someone with a lot of time
and patience.  We could make a branch for this if somebody is serious
about working on it.

Tom