[00:27:52] TimStarling, it seems like LocalFile::purgeHistory doesn't actually do anything useful [00:28:43] it's like it expected getCacheKey() to be overridden in OldLocalFile to use oldfile instead of file [00:28:48] but it actually just returns false [00:30:08] maybe it did something when it was introduced, but was not updated later [00:30:35] if there's no cache to clear, presumably the right thing to do is to remove it [00:30:40] yep [00:48:49] TimStarling: how does the table produced by https://gerrit.wikimedia.org/r/#/c/206740/1/cli/wm-scripts/bv2015/populateEditCount.php get used? [00:49:12] s/produced/populated/ [00:49:39] that's the other change, that I just finished [00:49:43] https://gerrit.wikimedia.org/r/#/c/206999/1/cli/wm-scripts/bv2015/voterList.php,unified [00:49:55] oh ok [00:49:57] * ori reads [00:50:01] * AaronSchulz wonders what the wave of INSERTS on https://tendril.wikimedia.org/host/view/db1072.eqiad.wmnet/3306 was about [00:50:20] the table created by populateEditCount.php is used by voterList.php [00:50:58] then the table populated by voterList.php is used by SecurePoll_LocalAuth [00:55:15] AaronSchulz: that was presumably voterList.php [00:55:44] adding 25M rows to bv2015_edits [05:35:04] TimStarling: when are you hoping to have that patch merged? It LGTM. I'm going to run it in beta, since it is sufficiently like production for this purpose. [05:35:39] Brad will probably find something I missed if, if you can wait for him to review it. [05:35:46] I was going to run it as soon as populateEditCount.php finishes [05:36:02] you can merge it now if you like [05:36:25] OK. [05:36:58] I ran each DB cluster in parallel, I'm just waiting on s3 now [05:37:15] it got hung up on loginwiki overnight, I killed that one this morning [05:40:01] i hope james & philippe realize you basically saved their asses here and i hope they plan on approaching this with a little more forethought next time around [13:28:34] ori: I didn't see anything worth mentioning [14:55:25] ^d: can you review https://github.com/wikimedia/cdb/pull/1 ? (needs to be pulled in to gerrit if it's good) [14:57:01] <^d> Heh, I'm hoping those tests cover the PHP implementation :p [16:23:30] legoktm, https://gerrit.wikimedia.org/r/#/c/206879/ [16:41:22] legoktm: can we mark https://phabricator.wikimedia.org/T961 as {{done}} ? [16:59:44] bd808: closed [17:10:06] legoktm: is there any SULF stuff that you need help? [17:10:45] not right now, I might have some non-urgent patches later today [17:11:20] k. I'm trying to figure out where to start being a developer again. :) [17:22:41] <^d> bd808: https://phabricator.wikimedia.org/maniphest/?statuses=open%2Cstalled&allProjects=PHID-PROJ-tjayastrttzou3qlleru#R [17:22:43] <^d> Start there ;-) [17:23:52] ^d: how does that relate to Reading ;) [17:24:14] <^d> maybe some of the bugs do? [17:24:14] <^d> :p [17:24:28] * bd808 sees a bunch of document logging stuff that he should do [17:24:30] docs need to be up to date so people can read them! [17:24:36] perfect! [17:25:35] <^d> Also T86363 :p [17:58:55] ^d: that's easy, just make anything sans-serif except English wikis ^^ [17:59:19] <^d> Easy? Write a patch and get it merged :) [17:59:48] API team: room isn't yet cleared out, so might not start quite on time [18:00:36] ^d: well, is editing this file the correct way to do that? https://gerrit.wikimedia.org/r/#/c/200940/1/resources/src/mediawiki.legacy/shared.css,cm [18:02:08] <^d> I dunno :p [18:08:34] <^d> AaronSchulz: Can you remove me from https://wikitech.wikimedia.org/w/index.php?title=Special:NovaProject&action=deletemember&projectname=wikidata-dev ? [18:12:10] bd808: Am I missing anything important at the meeting? This diner failed to have wifi but you can add my cell phone if necessary [18:12:42] marktraceur: we're just talking about stuff we're working on [18:12:53] and if we need any help [18:12:55] marktraceur: right, each person is giving status [18:13:11] (of api team stuff) [18:13:13] Okay. I won't worry about it too much...I'm pretty much shifting anyway [18:13:39] I need review on some stuff and I will circle back to address review every so often [18:13:43] just make sure bd knows the state of anything you're leaving behind [18:15:08] KK [18:15:55] marktraceur: I started looking for your patches needing review [18:16:01] bd808: Everything is terrible, but slightly less terrible than it was before, and anomie needs to review stuff and I need to fix stuff. {{done}} [18:16:09] perfect [18:16:21] OK cool. It's not a lot of patches, but some are big [18:17:11] With that, I return to internetless land. [18:20:21] bd808: wanna work on https://phabricator.wikimedia.org/T97446 ? :) [18:23:47] legoktm: ooooh something I understand! [18:26:28] ^d, done [18:26:38] <^d> thx [20:36:19] legoktm, bd808, ^d, AaronSchulz: Anyone want to review https://gerrit.wikimedia.org/r/#/c/207272/ quick, to help stop a flood of errors from Commons? [20:37:10] greg-g: Any opposition to me deploying https://gerrit.wikimedia.org/r/#/c/207272/ as soon as someone merges it? Tons of errors on Commons. [20:40:06] anomie: looking [20:44:15] anomie: I don't quite follow the why this fixes the error bit, but the code look solid. Can +2 based on "Brad says this is the right fix" [20:46:42] bd808: The error is because the thing is returning arrays that include "_type" as a key, while the API now reserves keys beginning with "_" for metadata. In particular, "_type" is the internal value for ApiResult::META_TYPE, so it's throwing an error about the unrecognized value for that bit of metadata. The patch fixes it by setting a different bit of metadata to say that "_type" shouldn't be considered metadata in that particular place. [20:47:57] *nod* that makes sense [20:47:59] Thanks legoktm [20:48:08] +2 race! [20:59:47] anomie: no opposition [20:59:57] greg-g: Good, because I just did ;) [21:00:01] :) [21:00:11] Anyone know how to delete all memc keys with a particular prefix? [21:03:39] hahaha [21:04:49] <^d> anomie: Fetch all the keys, iterate, delete as needed. [21:04:54] <^d> Or delete all the keys [21:05:58] Anyone want to volunteer to do that for keys starting with "commonswiki:getExtendedMetadata:"? [21:07:31] <^d> Which actually we don't have an accessor to that in MemcachedBagOStuff [21:08:20] keys() blocks the whole server...not sure for how long [21:08:59] <^d> We have a keys() function? [21:09:59] <^d> Oh calling it in memc, nevermind [21:10:57] easier to change the code to add v1 or something [21:11:13] <^d> That too [21:13:44] so...add a wmf hack for the next 30 days to add "v1" to the keys somewhere? [21:14:12] For that matter, we could add a hook for 'ValidateExtendedMetadataCache' that rejects anything before the appropriate timestamp. [21:20:48] legoktm, AaronSchulz, ^d, bd808: Any volunteers to merge and deploy https://gerrit.wikimedia.org/r/#/c/207279/? [21:22:41] Test case: https://commons.wikimedia.org/w/api.php?titles=File:Meganyctiphanes_norvegica2.jpg&iiprop=extmetadata&prop=imageinfo&action=query&iiextmetadatamultilang=1 [21:23:39] anomie: I +1'd it, but you should probably be the one to merge + deploy it [21:23:52] legoktm: I just noticed I had the test backwards :/ [21:23:58] can it just go in swat? [21:24:06] awkward >.> [21:24:37] bd808: Could, if someone wants to be there for it (I won't be around) and if we don't mind stuff on Commons being broken for another 1.5 hours. [21:25:24] I have time, I can deploy it [21:25:29] <^d> oh look [21:25:33] <^d> it got a +2 [21:25:37] <^d> better deploy ;-) [21:26:00] <^d> Would've been more funny if I hadn't gotten caught behind a mw/core job [21:27:14] * bd808 logs into tin to do ^d's bidding [21:27:19] <^d> I'm already in [21:27:26] sweet [21:27:55] Ok, I'll let ^d do it. [21:28:23] <^d> Hey, bd808 said he wanted me to more like Reedy :p [21:28:25] <^d> jfdi! [21:29:14] Reedy was our chaosmoney. kept everyone on their toes [21:29:26] and resolved a zillion bugs in the process [22:28:06] legoktm: something about the api format changes in 1.26wmf3 is breaking stuff for analytics [22:28:16] uhoh [22:28:23] is there a specific API module? [22:28:27] this -- https://meta.wikimedia.org/w/api.php?action=jsonschema&title=MobileWikiAppArticleSuggestions&revid=11448426 [22:28:38] the "required"s should be bools [22:28:52] ok, it was probably outputting raw "true" stuff [22:28:53] schema is here -- https://meta.wikimedia.org/wiki/Schema:MobileWikiAppArticleSuggestions [22:29:21] I figured the fix would be easy if we know where to look [22:29:55] yeah it should be easy, we just fixed this in TemplateData a few days ago [22:30:24] is it just "required"? or are there other properties? [22:31:36] mforns only pointed out required as being a problem. [22:31:49] he's in #wikimedia-analytics