> Thomas Lockhart <lockhart(at)alumni(dot)caltech(dot)edu> writes:
> > One hesitation I have is the performance hit in mixing FLOAT and
> > NUMERIC; I (probably) don't want to make NUMERIC the "best" numeric
> > type, since it is potentially so slow.
>
> I concur --- I'd be inclined to leave FLOAT8 as the top of the
> hierarchy. But NUMERIC could be stuck in there between int and float,
> no? (int-vs-numeric ops certainly must be promoted to numeric...)
If you cast NUMERIC to FLOAT8, then you would loose precision and it would
be counterintuitive type promotion (at least for a C programmer). If someone
wants speed over correctness, he can always explicitly cast NUMERIC to
FLOAT8. Seems like "correct" should take precedence over "fast", at least as
long as there is a way to do "fast".
Gene Sokolov.