[v3] PATCH to revert the aliasing problem workaround in function::swap

[Jason Merrill]

On 05/27/2010 01:44 PM, Paolo Carlini wrote:
> On 05/27/2010 06:42 PM, Jason Merrill wrote:
>> With recent improvements in alias analysis, this workaround is no
>> longer necessary; I'm restoring the original code that just does union
>> assignment so that the testsuite will complain if we start making
>> invalid assumptions again.

> That's good to know, thanks.
>
> Thus, it means that both 42832 and 42845 can be just closed?

I think so, yes.

Jason