Re: pg_typeof() patch review

On Nov 3, 2008, at 10:02 AM, Tom Lane wrote:

> "David E. Wheeler" <david@kineticode.com> writes:
>> On Nov 3, 2008, at 1:28 AM, Kurt Harriman wrote:
>>> 2) func.sgml: clarifying that the function returns an OID rather
>>> than a string
>
>> Actually, it returns a regtype, no?
>
> I thought the description was good, because it emphasizes that the
> result is-a OID; the table entry says "regtype" but people might not
> realize that that means they can use it as, eg, something to compare
> to pg_attribute.atttypid.

Well, as someone who was until recently unfamiliar with regtypes, and
who thinks of an OID as essentially just a number, I would find it
very useful if the description indicated that, as a regtype, the
return value could be used as either an OID or as string. Otherwise,
I'd find the description kind of confusing (in one place it says it
returns a regtype, "whatever *that* is", and in one place it says an
OID).

Just thinking at this from the point of view of a relative newbiew…

Thanks,

David