This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: PR c++/5504: Optimization breaks wei-ku-1 from blitz
- From: Mark Mitchell <mark at codesourcery dot com>
- To: Jason Merrill <jason at redhat dot com>, Richard Henderson <rth at redhat dot com>, Diego Novillo <dnovillo at redhat dot com>
- Cc: "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>, "gcc-gnats at gcc dot gnu dot org" <gcc-gnats at gcc dot gnu dot org>
- Date: Sat, 06 Apr 2002 14:48:54 -0800
- Subject: Re: PR c++/5504: Optimization breaks wei-ku-1 from blitz
- References: <wvlhempwz9b.fsf@prospero.cambridge.redhat.com>
> For 3.1, one option would be to go back to doing destructor cleanups only
> on the normal flow path; this would break EH semantics again for this
> case, but that would not be a regression and I don't think I've ever seen
> a bug report about it in past releases.
I think this is the most practical solution under the circumstances.
At some point, we clearly need to explore the other alternatives you
suggest.
Thanks,
--
Mark Mitchell mark@codesourcery.com
CodeSourcery, LLC http://www.codesourcery.com