Re: RFC: next/finish/etc -vs- exceptions

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:

Joel> Actually, we have two possible mechanisms, but I think we prefer the
Joel> "Zero Cost Exception" mechanism, which doesn't use setjmp/longjmp
Joel> because having an exception handler then costs nothing. We assume
Joel> that exceptions are rare occurences, and therefore don't wan't to
Joel> spent time setting up a handler if we're not going to use it.

Ok, great.  If Ada uses the same unwinder code as the rest of gcc --
in particular, unwind-dw2.c -- then this patch will "just work" for
Ada as it does for Java and C++.  That is because the debug hook in
the unwinder is inserted at the very lowest level.

Note that I have not updated all the unwinders.  In particular this
will not work on IA-64, on any platform using unwind-sjlj.c (but those
should be covered by the ordinary longjmp code), or I think on Xtensa
or ARM (based on the existence of unwind-$arch.c in the gcc tree).

Tom