[v3] fix libstdc++/51083

[Jonathan Wakely]

On 12 November 2011 16:24, Paolo Carlini wrote:
> Hi
>
>> I'm undecided whether this counts as a regression or not and if it
>> should be applied to the 4.6 branch - the testcases worked before TR1
>> was added or without -std=c++0x.
>
> Thanks a lot for fixing this! About the branch, the patch seems safe enough to me, maybe just wait a couple if weeks?

Yeah, that's what I was thinking too - will do, thanks.