Re: [RFA] Revised C++ ABI abstraction patches

[Jim Blandy <jimb at zwingli dot cygnus dot com>]

Michael Elizabeth Chastain <chastain@cygnus.com> writes:
> Lightly proofread, no obvious errors found.

Great.  That means all the bugs still there are non-obvious, and will
take a long time to find.  Just make my day, will you.  (Thanks.)

> I would prefer constructor_prefix_p et al to be real functions rather
> than macros.  When I single-step code in gdb, it's a lot easier to follow
> real functions than macros.

I agree completely.  But it is not the standard practice (see target.h
and elsewhere) and others don't seem to mind.  So I reined in my prima
donna urges and left them as macros.

And besides, function calls are so slow.  Remember, GDB's performance
matters a lot --- it's used to debug real-time operating systems!

:(

> If I get some bandwidth I'll test it against gcc 2.95.2 and gcc-3pre
> on a Solaris machine.

If I get a bandstand I'll play the boogie 'til the sun goes down.