[01:23:59] <Krinkle>	 Amir1: from reading T373037, I believe you're essentially proposing to support hash stops in SqlBagOStuff, right?
[01:23:59] <stashbot>	 T373037: Make ParserCache more like a ring - https://phabricator.wikimedia.org/T373037
[01:28:38] <Krinkle>	 i.e. instead of `mswiki:pcache:idoptions:514481` and `mswiki:pcache:idhash:514481-0!canonical` we could have `mswiki:pcache:514481|#|metadata` and `mswiki:pcache:514481|#|output!canonical`
[06:03:51] <TimStarling>	 Krinkle: as Bartosz said, reset() is just a shortcut for extracting the first element of an array
[06:04:35] <TimStarling>	 arguably I could have used $errors[0], but that's just style
[06:06:16] <TimStarling>	 reset() was a defensive style I used in case of misnumbered arrays caused by unset()
[09:30:28] <Amir1>	 Krinkle: for the first part, I'm proposing to have a sharding key which would be the same for idoptions and idhash (only being set for wiki id + page_id) so when we remove a pc cluster/host, only 1/nth get affected
[14:25:18] <Krinkle>	 TimStarling: ah, of course. I've actually seen us do that before, right. First internally ordered key regardless of sequential or assoc. I forgot about that part of reset(). That makes a lot more sense.
[14:25:28] <Krinkle>	 value*