Re: [sqlite] Speeding up slow select

2013-07-11 Thread GB
The column status_timeline_relationship.status_id is not indexed so sqlite has no way for a speedy lookup of rows by status_id. Thus it looks up all matching rows from status_timeline_relationship and does index-based lookups in status. Try an index on status_id or on (status_id, timeline_id).

Re: [sqlite] Speeding up slow select

2013-07-10 Thread Simon Slavin
On 11 Jul 2013, at 3:17am, Igor Tandetnik wrote: > SQLite believes that the fastest way is to start by finding > status_timeline_relationship records with timeline_id = 2, then join those > back to status and sort the result. > > It seems that

Re: [sqlite] Speeding up slow select

2013-07-10 Thread Igor Tandetnik
On 7/10/2013 9:56 PM, Tyler Spivey wrote: I'm trying to speed up this query, and don't understand why it's not using ix_status_created_at_sort. created_at_sort is a sorted column I'm using as part of a scrolling cursor for moving forward/backward through results, and the status table has

[sqlite] Speeding up slow select

2013-07-10 Thread Tyler Spivey
I'm trying to speed up this query, and don't understand why it's not using ix_status_created_at_sort. created_at_sort is a sorted column I'm using as part of a scrolling cursor for moving forward/backward through results, and the status table has ~36000 rows. SELECT status.text FROM status JOIN