--- Comment #11 from Aryeh Gregor <> 2011-02-25 
18:16:39 UTC ---
Since Delta was asking about this in #mediawiki, a summary:

1) The current query without namespace filtering will never scan more rows than
the number of results it returns, e.g., 500.

2) The current query with namespace filtering will scan millions of rows in the
worst case.

3) There is no likely way to change the schema to make the queries efficient. 
The only way I can think of is to denormalize unacceptably, as Brion says in
comment 1.

4) For some reason, the query is enabled anyway for the API.  I guess we could
disable that too, but it's not a reason to enable it in the web UI.  We don't
like queries that scan millions of rows worst-case.

Delta tells me, though, that the thing people really want this for is to limit
to the main namespace, or maybe to content namespaces.  This is much less of a
problem in practice than arbitrary namespace limitations.  The real issue would
be someone searching for all * links in Portal_talk or something.
 For real-world queries, limiting to content namespaces should only increase
rows scanned by a fairly small fixed factor, probably less than ten.  So that
can be considered, IMO.

Configure bugmail:
------- You are receiving this mail because: -------
You are on the CC list for the bug.

Wikibugs-l mailing list

Reply via email to