Re: floating point representation

Pete Forman <pete(dot)forman(at)westerngeco(dot)com> writes:
> Please remind me what we are trying to do.

The real point is that we need to serve several different purposes
that aren't necessarily fully compatible.

The existing default of FLT_DIG or DBL_DIG digits seems like a good
general-purpose policy, but it doesn't meet all needs. For pg_dump,
we clearly would like to promise exact dump and restore. On the
other side, the geometry regress tests would like to suppress a few
of the noisier low-order digits. And we frequently see questions from
users about how they can display fewer digits than the system wants to
give them --- or, more generally, format the output in some special
form.

I think the idea of making a user-settable format string is a good one.
I'm just afraid of the idea of trying to shoehorn in a solution at the
last minute; if we do, we may find it's not quite right and then have
a backwards-compatibility problem with fixing it. Besides, we are in
"no new features" mode during beta. I think it should wait for 7.2.

regards, tom lane