https://bugzilla.wikimedia.org/show_bug.cgi?id=29287

--- Comment #4 from Reedy <s...@reedyboy.net> 2011-06-06 19:02:01 UTC ---
Basic api module added in r89588

--- Comment #5 from Brion Vibber <br...@wikimedia.org> 2011-06-06 19:02:01 UTC 
---
Those row counts are estimates to begin with, and the difference is pretty tiny
(636k rows vs 658k rows? same order magnitude). I wouldn't assume that to help
all by itself...

However it looks like the extension doesn't actually add a 'repo_path' key.

The primary key as currently defined consists of (cp_repo_id, cp_rev_id,
cp_path).

cp_repo_id is a constant (within the given repository we're working with), but
cp_rev_id will vary. If we're returning filtered results listed by time -- as
we are for what you actually type in -- then that's pretty good. But if we're
trying to let someone type an arbitrary path, we probably want to bias to the
*path prefix* not to *recent revision ids*.

Searching based on an index built on (cp_repo_id, cp_path) will give more
efficient path lookups, but probably isn't super efficient either because it'll
have a huge number of duplicate (for our purposes) rows listing the same paths
over and over.

We probably don't want the DB to go through all 600k+ matching rows building a
DISTINCT set (which it might, or might not, be smart enough to optimize out).

-- 
Configure bugmail: https://bugzilla.wikimedia.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
You are on the CC list for the bug.

_______________________________________________
Wikibugs-l mailing list
Wikibugs-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikibugs-l

Reply via email to