zeitgeist team mailing list archive

[Markus Korn <thekorn@xxxxxx>]

** Branch linked: lp:~thekorn/zeitgeist/db_schema_3

** Branch unlinked: lp:~thekorn/zeitgeist/faster_origin_queries

** Changed in: zeitgeist
     Assignee: (unassigned) => Markus Korn (thekorn)

** Changed in: zeitgeist
       Status: Triaged => In Progress

-- 
Using the subj_origin column of event_view is slower than it should be
https://bugs.launchpad.net/bugs/673452
You received this bug notification because you are a member of Zeitgeist
Framework Team, which is subscribed to Zeitgeist Framework.

Status in Zeitgeist Framework: In Progress

Bug description:
Let me summarize what we have wrt subj_origin
 1) we have a subj_origin column in the event table, which is an integer pointing to an entry in the uri table
 2) we have an index on the subj_origin column of the event table
 3) we have subj_origin in event view, which is not the id, but the resolved uri 
 4) we have subj_origin_id in event_view which is in fact the same as event.subj_uri, but we get this value by using a sub-query on the uri table.

Points 3) and 4) are the reason why the index on event.subj_origin is never used when querying the event_view, example:

DEBUG:zeitgeist.sql:Got query:
QUERY:
SELECT id FROM event_view WHERE subj_origin_id=1 ([])
PLAN:
[0, 0, u'TABLE event']
[0, 0, u'TABLE uri USING PRIMARY KEY']

By simply renaming "event_view.subj_origin" to "event_view.subj_origin_uri" and "event_view.subj_origin_id" to "event_view.subj_origin" in combination with removing the sub-query we can make queries for the origin id on event_view *much* faster.