This is the mail archive of the
mailing list for the GCC project.
Re: C++/6527: destructors called in wrong order: regression from 3.0.x
- From: Janis Johnson <janis187 at us dot ibm dot com>
- To: Paolo Carlini <pcarlini at unitus dot it>
- Cc: Mark Mitchell <mark at codesourcery dot com>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>, janis187 <janis187 at us dot ibm dot com>
- Date: Wed, 1 May 2002 16:25:16 -0700
- Subject: Re: C++/6527: destructors called in wrong order: regression from 3.0.x
- References: <3CD060E5.email@example.com> <firstname.lastname@example.org> <3CD06B15.email@example.com>
On Thu, May 02, 2002 at 12:24:21AM +0200, Paolo Carlini wrote:
> Mark Mitchell wrote:
> > 2002-04-23 Jason Merrill <firstname.lastname@example.org>
> > PR c++/5504 - but re-breaks c++/411
> > * init.c (push_base_cleanups): Rename to perform_base_cleanups.
> > Expand base cleanups now rather than pushing them.
> > * decl.c (begin_destructor_body): Don't call push_base_cleanups.
> > (finish_destructor_body): Call perform_base_cleanups.
> > * cp-tree.h: Adjust prototype.
> > Would you try reverting this patch and see what happens? That's
> > one thing that's been in this area recently.
> Ok. I'm rebuilding now.
> If that reversion is the culprit, however...
I tried it, too, and it is the culprit. 3.1 CVS sources from this
afternoon fail the test, and with this patch reverted the test passes.
Now I can go back to finding out what broke Boost.