On Mon, Dec 05, 2005 at 08:24:16AM -0500, John D. Burger wrote:
> >>There are practical applications, eg, 1024-bit keys are fairly common
> >>objects in cryptography these days, and that equates to about 10^308.
> >>I don't really foresee anyone trying to run crypto algorithms with SQL
> >>NUMERIC arithmetic, though ...
> >
> >2046 bit keys are becoming more common. However, math using these keys
> >is
> >usually done modulo a product of two primes and there are ways of
> >doing the
> >calculations that are going to be much faster than doing them the way
> >Postgres does. So it is unlikely that anyone would be using Postgres'
> >numeric
> >type to do this in any case.
>
> Nonetheless, the fact that people can think of practical applications
> for numbers whose length is easily within a factor of two of the
> proposed limitation makes me squeamish about it being shrunk. Also, I
> would say the same arguments about doing math with NUMERICs suggest
> that saving a few byes in representation is not a big deal. On the few
> occasions where I have used NUMERICs, I didn't care about stuff like
> that.
I think that if there are any esoteric cases where people are doing
these kinds of things with numeric, they could probably be best answered
by offering a completely different system anyway, using a different type
name. The 5 people in the world doing this will just have to change
their code I guess... ;)
--
Jim C. Nasby, Sr. Engineering Consultant jnasby(at)pervasive(dot)com
Pervasive Software http://pervasive.com work: 512-231-6117
vcard: http://jim.nasby.net/pervasive.vcf cell: 512-569-9461