Re: Finer Extension dependencies

Tom Lane <tgl@sss.pgh.pa.us> writes:
>> That's how I did it first, but Alvaro opposed to that because it allows
>> for more than one extension to provide for the same feature name.
>>   http://archives.postgresql.org/pgsql-hackers/2012-03/msg01425.php
>
> Right, but the question that has to be considered is how often would
> that be intentional as opposed to an undesirable name collision.
> I think Hitoshi was right upthread that it will seldom if ever be
> the case that somebody is independently reimplementing somebody
> else's API, so the use-case for intentional substitution seems thin.

I reverted that change and we're now back to:

Table "pg_catalog.pg_extension_feature"
   Column   | Type | Modifiers
------------+------+-----------
 extoid     | oid  | not null
 extfeature | name | not null
Indexes:
    "pg_extension_feature_index" UNIQUE, btree (extoid, extfeature)
    "pg_extension_feature_oid_index" UNIQUE, btree (oid)

Regards,
--
Dimitri Fontaine
http://2ndQuadrant.fr     PostgreSQL : Expertise, Formation et Support