Kless wrote:
> I write here the answer of Jerry Stuckle [1] because it looks me
> interesting and enough logical.
>
>
> [1] http://groups.google.com/group/comp.databases.mysql/browse_thread/thread/89557609239a995e
> -----------------------
> Quite frankly, I don't care that PostGres has user-defined types.
> They
> restrict you to a single database, when others might be better for
> other reasons.
>
> And yes, I think other things should have been proposed to the SQL
> standards committee. It doesn't take that long to get a good proposal
> into the standards. No, it isn't immediate. But if there is a case
> to
> be made for it, then the committee will act.
>
> Then all databases get the feature, eventually.
>
> As I said. Do it the right way. Submit your proposal. If you have a
> case, it will be added to the SQL standard. If not, then it's not
> that
> important.
>
>
The time taken to get something into the standard is a lifetime in
computing terms. If my client has a need for a UDF they need it now, not
when the standards committee gets around to thinking about it.
Many UDTs will be specialised to a single user, and never be candidates
for inclusion in the standards. Excluding any type that isn't in the
standard would be to throw away one of Postgres' greatest strengths, one
we are justly proud of. Maybe you don't care about that, but we do, and
our clients do.
In any case, a standard for UUIDs would almost certainly not specify how
it is to be stored, which is where we got into this discussion.
This debate started with a misconception about how Postgres actually
stores UUIDs, and doesn't seem to have gained much point since then.
cheers
andrew