Re: Question about RUM-index

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

On Wed, Jun 15, 2016 at 3:56 AM, Andreas Joseph Krogh <andreas@visena.com> 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?

 

Note that GIN does almost what I want, except use the index when sorting by "sent"-timestamp.

 

So I wonder if RUM can do any better?

What I don't understand is how to have "folder_id" as part of the RUM-index so that I can search in an array of folders using the index, AND have the whole result sorted by "sent"-timestamp also using the RUM-index.

 

I think you would have to implement an operator for integers for RUM much like btree_gin does for GIN.  Sorry don't know how to do that, except to say look in the RUM code to see how it does it for time-stamps.

 

 

In the (limited) documentation sorting using timestamp is done like this:

 

ORDER BY sent <-> '2000-01-01'::TIMESTAMP

 

which I don't understand; Why must one specify a value here, and how does that value affect the result?

 

 

This is essentially identical to ORDER BY ABS(sent - '2000-01-01'::TIMESTAMP);  except it can use the index.

 

So maybe pick a constant outside the range of possible values, and use that as one argument to <->.

 

This should be unnecessary and hidden from the user. Maybe some "ORDER BY rum_timestamp(sent)" or something could abstract away stuff to make it much clearer to the user?

--

Andreas Joseph Krogh

CTO / Partner - Visena AS

Mobile: +47 909 56 963

andreas@visena.com

www.visena.com