This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Solving mt_allocator's static initialization ordering problem
- From: Brad Spencer <spencer at infointeractive dot com>
- To: Benjamin Kosnik <bkoz at redhat dot com>
- Cc: pcarlini at suse dot de, libstdc++ at gcc dot gnu dot org
- Date: Fri, 2 Jul 2004 21:15:30 -0300
- Subject: Re: Solving mt_allocator's static initialization ordering problem
- References: <20040506151114.GR21258@eng.infointeractive.com> <20040628173638.GO12291@eng.infointeractive.com> <40E2A433.5050908@suse.de> <20040630150654.GB3159@eng.infointeractive.com> <20040630213247.3315e581.bkoz@redhat.com>
On Wed, Jun 30, 2004 at 09:32:47PM -0500, Benjamin Kosnik wrote:
>
> Dudes, excellent work.
>
> This patch is in. For the first time, darwin results with mt_alloc match
> __pool_alloc and new_alloc. Whooo hoooo!!!!
Excellent :)
> I'm writing docs for __mt_alloc::_Tune.
>
> Let's switch the linux defaults now. Let me test a bit more on 64bit, non-x86 linux.
>
> tested x86/linux
> tested powerpc/darwin
I ran the libstdc++ tests with this on the 3.4 maint branch on
sparc-sun-solaris2.8 and everything passed except for two seemingly
unrelated tests. Maybe the patch can go in for 3.4.1? I'd also be
comfortable with this as the default allocator on Solaris. It's
definitely better than malloc. :)
The failures seem to be about iterator conversions. I'll look into
them when I get a chance.
FAIL: 23_containers/multiset/14340.cc (test for excess errors)
FAIL: 23_containers/set/14340.cc (test for excess errors)
--
------------------------------------------------------------------
Brad Spencer - spencer@infointeractive.com - "It's quite nice..."
Systems Architect | InfoInterActive Corp. | A Canadian AOL Company