[01:44:54] <bawolff>	 So in mysql, can ICP use indexes other than the index being used for the query?
[01:45:36] <bawolff>	 For context, I'm using the query: SELECT cl_from FROM categorylinks where cl_timestamp > '2013-07-03 4:15:21' and cl_to = "Noindexed_pages" and cl_type <> 'subcat' order by cl_timestamp limit 5000
[01:45:54] <bawolff>	 And I'm trying to figure out how ICP seems to apply to the cl_type part
[01:46:28] <bawolff>	 And why the explain seems to say filesort, eventhough in practise it does not seem to filesort when I'm running it
[02:04:31] <bawolff>	 Even if it could do that, it seems like that would be inefficient
[02:05:15] <bawolff>	 Since it wouldn't know where in cl_sortkey index to look
[02:05:57] <bawolff>	 Maybe just having that extra condition confuses mariadb, and causes it to do ICP for cl_timestamp, even though it already knows cl_timestamp is in range due to the RANGE index access method
[02:06:14] <bawolff>	 This is in the context of T200703
[02:06:14] <stashbot>	 T200703: Special:RandomInCategory does not return all pages with equal probability - https://phabricator.wikimedia.org/T200703
[16:17:47] <notthere>	 hello
[16:18:42] <Urbanecm>	 hi notthere 
[16:20:30] <notthere>	 what is this irc for?
[16:20:50] <Urbanecm>	 notthere: as its topic says, for "technical help for Wikimedia wikis"
[16:20:56] <notthere>	 yeah i read that
[16:21:08] <Urbanecm>	 usually, people from Wikimedia technical community come here to discuss things specific to Wikimedia
[16:21:16] <notthere>	 ohhhh ok
[16:21:20] <notthere>	 well ill be off then
[16:21:31] <Urbanecm>	 feel free to idle here if you want notthere 
[16:21:40] <Urbanecm>	 general conversation about development happens in #wikimedia-dev and #mediawiki
[16:21:50] <notthere>	 ok