Re: Reduce NUMERIC size by 2 bytes, reduce max length to 508

Tom Lane wrote:
> "John D. Burger" <john(at)mitre(dot)org> writes:
> > Tom Lane wrote:
> >> Hm ... between that, the possible crypto connection, and John's
> >> personal testimony
>
> > Just to be clear, this John has yet to use NUMERIC for any
> > calculations, let alone in that range.
>
> My mistake, got confused as to who had said what.
>
> The point remains though: in discussing this proposed patch, we were
> assuming that 10^508 would still be far beyond what people actually
> needed. Even one or two reports from the list membership of actual
> use of larger values casts a pretty big shadow on that assumption.

Agreed. I would like to see us hit the big savings first, like merging
cmin/cmax (4 bytes per row) and reducing the varlena header size (2-3
bytes for short values), before we start going after disk savings that
actually limit our capabilites.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073