On Wed, Mar 8, 2017 at 7:00 PM, Stephen Frost <sfr...@snowman.net> wrote: > Ashutosh, > > * Ashutosh Bapat (ashutosh.ba...@enterprisedb.com) wrote: >> On Mon, Mar 6, 2017 at 7:54 PM, Amos Bird <amosb...@gmail.com> wrote: >> > Well, the prefix is used to differentiate other \d commands, like >> > this, >> > >> > amos=# \ditv >> > List of relations >> > Schema | Name | Type | Owner | Table >> > --------+--------------------+--------------+-------+--------- >> > public | i | table | amos | >> > public | ii | index: gist | amos | i >> > public | j | table | amos | >> > public | jj | index: gin | amos | i >> > public | jp | index: btree | amos | i >> > public | js | index: brin | amos | i >> > public | numbers | table | amos | >> > public | numbers_mod2 | index: gin | amos | numbers >> > public | numbers_mod2_btree | index: btree | amos | numbers >> > public | ts | table | amos | >> > (10 rows) >> >> The header for this table is "list of relations", so type gets >> associated with relations indicated type of relation. btree: gin as a >> type of relation doesn't sound really great. > > The type is 'index', we're just adding a sub-type here to clarify the > kind of index it is. > >> Instead we might want to >> add another column "access method" and specify the access method used >> for that relation. But then only indexes seem to have access methods >> per pg_class.h. > > Right, I don't think having an extra column which is going to be NULL a > large amount of the time is good. The approach taken by Amos seems like > a good one to me, to have the type still be 'index' but with a sub-type > indicating the access method.
Ok. Seems like a good trade-off. -- Best Wishes, Ashutosh Bapat EnterpriseDB Corporation The Postgres Database Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
