On 07/09/2011 10:05 PM, Matthias Berndt wrote:
> I mailed Jason Merrill about this, as he wrote the patch that broke it. He
> told me he fixed it in revision 176087 of the 4.6 branch. Could you package
> that fix?

I would prefer if that would be tractable by an upstream report. I currently
can't see any connection about the upstream commit and the debian report :-/

Matthias



--
To UNSUBSCRIBE, email to debian-gcc-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4E18CDB1.9030203@debian.org">http://lists.debian.org/4E18CDB1.9030203@debian.org

> I would prefer if that would be tractable by an upstream report. I
> currently can't see any connection about the upstream commit and the
> debian report :-/
The commit adds a regression test which exhibits the exact same behaviour
as the test case I've sent you. It compiles fine without -std=c++0x, and if
you compile it with -std=c++0x, it refuses the program with the error
message "temporary of non-literal type ‘<whatever>’ in a constant
expression". This seems like a pretty strong connection to me.
Matthias



--
To UNSUBSCRIBE, email to debian-gcc-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 201107100644.59976.matthias_berndt@gmx.de">http://lists.debian.org/201107100644.59976.matthias_berndt@gmx.de