Thread: Patch to allow domains over composite types

typecmds.c says:
"Domains over composite types might be made to work in the future, but
not today."

Attached is a patch that allows domains over composite types, together
with test cases in domaincomp.sql. A domain over a composite type has
typtype TYPTYPE_DOMAIN, but typrelid and typrelkind are empty: that
information is only available in the pg_type record of the base type.
The remainder of the patch follows from that choice. While parsing a
record expression into a row type, an extra coercion node had to be
inserted to ensure that the domain checks are called.

All regression tests are ok, comments are highly appreciated.

--

Yeb Havinga
http://www.mgrid.net/
Mastering Medical Data

On Wed, May 11, 2011 at 8:07 AM, Yeb Havinga <yebhavinga@gmail.com> wrote:
> typecmds.c says:
> "Domains over composite types might be made to work in the future, but not
> today."
>
> Attached is a patch that allows domains over composite types, together with
> test cases in domaincomp.sql. A domain over a composite type has typtype
> TYPTYPE_DOMAIN, but typrelid and typrelkind are empty: that information is
> only available in the pg_type record of the base type. The remainder of the
> patch follows from that choice. While parsing a record expression into a row
> type, an extra coercion node had to be inserted to ensure that the domain
> checks are called.
>
> All regression tests are ok, comments are highly appreciated.

I don't have time to look at this right now, but please add it to the
currently-open CommitFest:

https://commitfest.postgresql.org/action/commitfest_view/open

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On 2012-02-27 12:49, Asif Rehman wrote:
> Hi Yeb Havinga,
>
> I was digging archives to see anyone worked on supporting domain's 
> over composite type and found your patch, but that was pulled back. 
> According to commitfest comments it needs some more work...

There were some issues with using the domains from pl/pgsql, which could 
probably made to work, but I didn't investigate it because at the time 
the use case for which is was needed was solved in a different way.
>
> Are you going to submit the updated patch?

There is no updated patch, sorry.

regards,
Yeb