Tom Lane wrote:
and unless I'm mistaken, our behavior conforms to the spec and Oracle's doesn't. Strictly speaking, the spec doesn't define the behavior of "SUBSTR" at all, only "SUBSTRING" with this weird FROM/FOR argument syntax. But PG treats SUBSTR(x,y,z), SUBSTRING(x,y,z) and SUBSTRING(x FROM y FOR z) all the same. Possibly Oracle conforms to spec for SUBSTRING but their SUBSTR acts differently?
Thanks, Tom, I agree that PG's substr() appears to be following the spec (thank goodness I don't have to read that whole thing ;). Oracle does not implement substring() at all (up through release 9.2, the latest I have to work with. Just checked the online documentation for 10g Release 2 (latest available) and it doesn't have it either. Says this under conformance: "E021-06, SUBSTRING function: use SUBSTR function instead". Sigh...
-- Guy Rouillier
