Re: exposing float8-as-value to PGXS/makefiles

Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> I was just messing around with a datatype that's based in int64
> representation.  Pretty much everything (seems to) work cleanly, but one
> problem I have is that I cannot build the correct CREATE TYPE sentence
> in the .sql.in file to actually install the type, because there's no
> easy way to figure out whether float64 (and therefore int64) is passed
> by value or not.

We already solved that for contrib/isn --- use the LIKE clause in CREATE
TYPE.
        regards, tom lane