Michael Paquier <michael.paquier@gmail.com> writes:
> On Wed, Mar 25, 2015 at 8:18 PM, Pedro Gimeno
> <pgsql-004@personal.formauri.es> wrote:
>> The only point of using copysign there was to deal with a negative zero
>> corner case. The attached removes copysign from the test program too,
>> relying instead on sprintf outputting the sign of minus zero when available.
> Cool, thanks. Applied to the code tree of Postgres, it gives the patch attached.
This is pretty desperately in need of some comments, but I'll see what
I can do with it.
I assume the consensus is that we should not back-patch this? It seems
like the kind of behavioral change that people might get annoyed about.
regards, tom lane