Re: Question about RUM-index

[Oleg Bartunov <obartunov@xxxxxxxxx>]

On Fri, Jun 17, 2016 at 4:50 PM, Andreas Joseph Krogh <andreas@xxxxxxxxxx> wrote:

På fredag 17. juni 2016 kl. 15:47:08, skrev Oleg Bartunov <obartunov@xxxxxxxxx>:

 

 

On Fri, Jun 17, 2016 at 3:52 PM, Andreas Joseph Krogh <andreas@xxxxxxxxxx> wrote:

På fredag 17. juni 2016 kl. 13:53:34, skrev Oleg Bartunov <obartunov@xxxxxxxxx>:

 

 

On Fri, Jun 17, 2016 at 2:10 PM, Oleg Bartunov <obartunov@xxxxxxxxx> wrote:

 

 

On Fri, Jun 17, 2016 at 9:32 AM, Andreas Joseph Krogh <andreas@xxxxxxxxxx> wrote:

På torsdag 16. juni 2016 kl. 00:50:45, skrev Jeff Janes <jeff.janes@xxxxxxxxx>:

On Wed, Jun 15, 2016 at 3:56 AM, Andreas Joseph Krogh <andreas@xxxxxxxxxx> wrote:

Hi.

 

First; Is this the correct forum to ask questions about the Postgres Pro's new RUM-index?

 

If not, please point me to the right forum.

 

I think that https://github.com/postgrespro/rum/issues might be the best forum.

 

Oleg and friends; Should we use GitHub-issues as forum (one issue per question/thread?), pgsql-general or something else?

 

Andreas,
 

we are hardly working on our internal version of rum and will open it after resolving some issues. I think the best place to discuss it is -hackers.

 

Ah, as someone corrected me, we are working hard !

 

He he, I figured that was what you meant:-)

 

Ok, so basically - will RUM-index support the same indexing-properties as GIN (being able to index tsvector, BIGINT-arrays, JSONB etc.) and be able to use index for sorting on ie. timestamp, tsrank or some BIGINT-column?

 

Like my example, will it be possible to issue a query like this:

 

SELECT del.id
                    , del.sent
                FROM delivery del
                WHERE 1 = 1
                      AND del.fts_all @@ to_tsquery('simple', 'hi:*')
                      AND del.folder_id = ANY(ARRAY[2,3]::BIGINT[])
                ORDER BY  del.sent DESC LIMIT 101 OFFSET 0;

 

and have it use one RUM-index so the whole operation, including sorting, is as efficient as possible?

 

 

we have implementation for timestamp. One need to write opclass to deal with arrays, it shouldn't be difficult.

 

Will the opclass dealing with bigint-arrays also handle the single-element case, that is only one bigint?

this is up to opclass author.

 

 

--

Andreas Joseph Krogh

CTO / Partner - Visena AS

Mobile: +47 909 56 963

andreas@xxxxxxxxxx

www.visena.com