Re: generic type support

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Øyvind" == Øyvind Harboe <oyvind dot harboe at zylin dot com> writes:

Øyvind> I was surprised to find that GCJ has a frontend at all.
Øyvind> Why? 

Øyvind> It seems like this problem has been thoroughly put to bed by e.g. 
Øyvind> Jikes, javac, etc.

It is a "natural" thing to do.  And it is certainly convenient, though
perhaps the convenience could be addressed by teaching gcj how to
invoke a bytecode compiler.  Maybe there are strange failure modes
involved in doing this.

Relying on javac was never an option (except at the very early days
when only the bytecode front end was working).  javac isn't free
software.

Were there free compilers around when gcj was started?  I don't
remember.

Anyway, more practically, the bytecode front end still generates worse
code.  I think we know how to fix it, but nobody is working on it.


Teaching gcj to invisibly run jikes and then read the resulting
bytecode would be an interesting project.

Tom