Re: Floating-point timestamps versus Range Types

On Mon, Oct 25, 2010 at 4:35 PM, James Cloos <cloos@jhcloos.com> wrote:
>>>>>> "TL" == Tom Lane <tgl@sss.pgh.pa.us> writes:
>
> JC> That said, the possiblity of hex i/o format for the float datatypes
> JC> would be welcome.
>
> TL> It's unportable, for two different reasons:
>
> TL> 2. The printf specifiers you want us to rely on are not standard.
>
> They are in C99.
>
> TL> 1. pg_dump output would become platform-specific.  This is highly
> TL> undesirable.
>
> It is true that pg would have to test for them in configure and supply
> alternative code wherever libc fails to support them.
>
> I can readily accept that there are many more pressing needs.
>
> But would such a patch for master be rejected?

Let me back up a few steps and ask why you want this in the first
place.  If there's a real problem here, we should solve it, either in
the way you've proposed or in some other manner.  But you haven't
really said what problem you're trying to solve - just that it would
be "welcome", which leaves much to my (not very good) imagination.

This is quite a bit OT for this thread so if you want to pursue this
I'd suggest starting a new thread in which you start by laying out
your case for doing this.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company