Re: Domain Support -- another round

New set with most of Peters comments corrected. Left the deal about
schema though :) Took nearly an hour to do a cvs diff for some reason
this time (normally a couple of minutes is enough).

> Random nitpicking below. Also, have you created a regression test?

They had been posted a few times and haven't changed. (Attached
anyway)

> > + <structfield>typnotnull</structfield> represents a NOT NULL
> > + constraint on a type. Normally used only for domains.
>
> And unnormally...?

Unnormally is when someone sets it by hand on a type which isn't a
domain -- I guess. Corrected.

> > + <!entity createDomain system "create_domain.sgml">
>
> I don't see this file included.

Other messages. Full package included on this one however.

> > + * MergeDomainAttributes
> > + * Returns a new table schema with the constraints, types,
and other
> > + * attributes of the domain resolved for fields using the
domain as
> > + * their type.
>
> I didn't know we had schemas yet. You should probably not overload
that
> term to mean "a list of database objects".

Merge attributes says something very similar about inheritance and
table schemas. Kinda correct considering
the variable used in both cases is *schema.

The diff weirdness in regards to DROP DATABASE is probably because I
started by copying the DROP DATABASE element, then altered it. I
don't know why it chose that method to do the diff though, but it is
accurate. Using -cd flags didn't make it any prettier.