Re: Updated RUM-index and support for bigint as part of index

På lørdag 06. august 2016 kl. 20:54:32, skrev Artur Zakirov <
a(dot)zakirov(at)postgrespro(dot)ru <mailto:a(dot)zakirov(at)postgrespro(dot)ru>>:
Hello,   2016-08-02 21:08 GMT+03:00 Andreas Joseph Krogh <andreas(at)visena(dot)com
<mailto:andreas(at)visena(dot)com>>:  The ORDER BY part seems strange; It seems one
has to find a value "lower than any other value" to use as a kind of base, why
is this necessary? It also seems that in order to be able to sort DESC one has
to provide a timestamp value "higher than any other value", is this correct?
 
It would be great if the docs explained this.
 
We will write more detailed documentation for RUM.

 
Great!
 
 
I really miss the opportunity to include a BIGINT as part of the index, so
that the WHERE-clause could be like this:
 
WHERE del.fts_all @@ to_tsquery('simple', 'andreas&kr') AND del.folder_id IN (1
,2,3)
 
Having this would be perfect for my use-case searching in email in folders,
sorted by received_date, and having it use ONE index.
 
Will this be supported?

We have a plan to use generic types to able to include bigint, timestamp and
other types as part of index.

 
Does this eliminate the need for a btree_rum equivalent of btree_gin, being
that the RUM-index will handle all "btree-able" datatypes?
 
 
But I cant tell date of it.

 
I understand.
Do you think it will be done by the time 9.6 is released?
 
Thanks.
 

-- Andreas Joseph Krogh
CTO / Partner - Visena AS
Mobile: +47 909 56 963
andreas(at)visena(dot)com <mailto:andreas(at)visena(dot)com>
www.visena.com <https://www.visena.com>
<https://www.visena.com>