This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: gcc 3.1: bug with inherited templates when specified by namespace
- From: Richard Smith <richard at ex-parrot dot com>
- To: Mark Mitchell <mark at codesourcery dot com>
- Cc: Gabriel Dos Reis <gdr at codesourcery dot com>, Dylan Cuthbert <dylan at q-games dot com>, "Mountfort, Geoff" <gmountfort at avaya dot com>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>
- Date: Thu, 27 Jun 2002 23:18:23 +0100 (BST)
- Subject: Re: gcc 3.1: bug with inherited templates when specified by namespace
On Thu, 27 Jun 2002, Mark Mitchell wrote:
> --On Thursday, June 27, 2002 07:07:00 PM +0200 Gabriel Dos Reis
> <gdr@codesourcery.com> wrote:
>
> > Richard Smith <richard@ex-parrot.com> writes:
> >
> > | On Mon, 24 Jun 2002, Dylan Cuthbert wrote:
> > |
> > | > Yep, you're getting the same problem then.
> > | >
> > | > I wonder what MrBlobbyBlab actually gets in his class in that first
> > example? |
> > | The base class gets ignored. This was reported as PR/3656 and fixed
> > | (together with PR/3123) on the main line in March with the patch
> > |
> > | 2002-03-21 Gabriel Dos Reis <gdr@merlin.codesourcery.com>
> > |
> > | PR C++/3656
> > | * semantics.c (finish_base_specifier): Handle erronous base
> > | classes.
> > |
> > | This patch has not been merged onto the 3.1 branch (although it probably
> > | ought to be) and both PR/3656 and PR/3123 still manifest there.
> >
> > Mark,
> >
> > In light of the above discussion, may I ask for approval to apply
> > the above mentioned patch to 3.1 branch?
>
> If the patch fixes regressions, yes. Otherwise, no.
Prior to 3.0 the use of a template-id or a namespace as a base class
caused an ICE, since then it has silently ignored the illegal base class.
I guess that going from ICEing on an invalid base class to silently
dropping the base class is a regression of sorts. At least with an ICE
it was clear there was an error of some sort, now it's not.
--
Richard Smith