Some initial diagnosis of memtemp89.C

[Paul Burchard]

Kriang Lerdsuwanakij wrote:
> I submitted two patches a few months ago that fix several template
> template parameter bugs.

Thanks!  These two patches together do indeed solve the test case
memtemp89.C (along with several related test cases that I didn't
submit).

>         http://egcs.cygnus.com/ml/gcc-patches/1999-04/msg00341.html

An improved version of the gcc/cp/pt.c part of this patch recently made
it in, as part of a patch for Koenig lookup by Alex Samuel.  Actually,
that was at least the third time that someone has offered a patch for
the 980422 ICE... but Alex's version does appear to be the "most
correct" of the three.  The demangle_template() part needs a bit of
freshening due to some Java mods to libiberty.

>         http://egcs.cygnus.com/ml/gcc-patches/1999-03/msg00000.html
...
> There was a change in the first patch that fixes the lookup_template_class()
> function call inside tsubst() you are investigating.

Right...except you better understood what was going wrong in that call
(your change clearly works, but I don't understand why it's the right
thing to do).

Anyway, if you don't do it, I will submit an updated version of your
patch once Mark Mitchell & Co. have finished completely rewriting gcc
for function-at-a-time processing. :-)  Mark is catching a lot of
template bugs in the process (though I doubt that your name mangling
fixes will come out of that).

-- 
----------------------------------------------------------------------
Paul Burchard   <burchard@pobox.com>   http://www.pobox.com/~burchard/
----------------------------------------------------------------------