Re: FAQ Update

[Tom Tromey <tromey at cygnus dot com>]

>>>>> "Jeff" == Jeff Sturm <jsturm@sigma6.com> writes:

Jeff> Are you using Alexandre's DECL_INLINE patches?  I saw this
Jeff> too... I worked around it by marking a few particular methods
Jeff> `final' instead of `private'... they are still inlined but not
Jeff> optimized away.

This is a problem.  In some cases we have native methods that use
private fields (or methods) in other packages -- so even a heuristic
like "don't remove an unused private method if there is a native
method" won't work right now.  (Probably we should change the code and
still use this heuristic, but I haven't really thought about it.)

Jeff> among them that jc1 can't inline native code and vice-versa.

In some cases g++ can inline Java code.  That's because gcjh makes an
attempt to decompile certain simple methods.  See decompile_method()
in gjavah.c.  We could do more here.  We could even do full
decompilation if we really wanted.

I agree it would be cooler if g++ and gcj could read each others' code
and really inline.  I doubt that will happen.

Tom