[03:32:30] TimStarling: you sound busy with the replication stuff, we can check in tomorrow or whatever :) [call is scheduled now] [03:33:13] we can do it now, if you're available [03:33:25] sure. call or irc? (irc fine by me) [03:34:05] alright, IRC [04:46:22] AaronSchulz: do you think you could take a look at https://gerrit.wikimedia.org/r/#/c/350958/ ? [05:30:47] TimStarling: Should mwconfig be visible at https://config-master.wikimedia.org/ ? [15:20:11] anomie: you around [15:20:12] ? [15:20:20] arlolra: Yes [15:20:42] can you help me out with something [15:20:44] in https://phabricator.wikimedia.org/T39902#2554331 [15:21:13] special/media don't seem to be in pagelinks [15:21:15] for example [15:21:16] https://en.wikipedia.org/w/api.php?action=query&prop=links&titles=User:Arlolra/sandbox [15:21:23] can you clarify what you meant there [15:28:39] arlolra: See for example https://en.wikipedia.org/w/api.php?action=query&titles=File:Example.svg, the page is missing on enwiki but it's a "known" page because it exists on Commons. Special and Media pages not being in pagelinks is a different task, which I can't find at the moment if I'm not misremembering that it even exists. [15:32:41] hmm [15:33:53] anomie: is there an efficient way to query for all the links on a page (including special, media, etc.) and return with missing/known? [15:33:57] VE seems to be doing this [15:33:58] https://github.com/wikimedia/mediawiki-extensions-VisualEditor/blob/HEAD/modules/ve-mw/init/ve.init.mw.LinkCache.js#L184-L195 [15:34:25] it gets a list of all the links on a page [15:34:25] https://github.com/wikimedia/mediawiki-extensions-VisualEditor/blob/HEAD/modules/ve-mw/init/ve.init.mw.LinkCache.js#L125 [15:34:40] and then submits those in succession [15:35:30] arlolra: If you have a list of titles, you can pass them to action=query to get missing/known. No prop module is needed, it's part of the basic action=query output. [15:35:50] how big a list can i pass? [15:35:56] hundreds? [15:36:19] But there's no way to get those links to Special and Media pages without parsing the wikitext yourself, since ParserOutput doesn't record them. [15:37:42] The auto-generated documentation tells you: "Maximum number of values is 50 (500 for bots)." action=paraminfo includes that information in a machine-readable format too. [15:48:47] i think the prop module is included because they also want to know about disambiguation pages [15:49:20] dismabig and redirects (adding mw-disambig and mw-redirect classes). [15:49:46] thanks, which i suppose we'll be adding as well [15:50:14] 500 for bots ... ok, so it sounds like that should be sufficient [15:50:40] parsoid can do the same thing VE is, collect all the links on a page, and pass that to parsoidbatchingapi so run the same query [15:51:20] arlolra: If you used the right coding standards you could do it with just a copy-and-paste! ;-) [15:52:29] you wound me [15:53:15] thanks for the help anomie [15:54:29] I wasn't saying VE didn't have a need for the prop modules it's using, sorry for the confusion. I was just saying that if you only care about the "missing" and "known" flags you don't need any prop modules to get them. [15:54:50] understood [16:44:37] James_F, the code is going to go into an existing PHP extension ... so ... [16:45:02] arlolra is feeling wounded unnecessarily :) [16:45:07] * James_F grins [18:34:07] tgr, utf8 is variable length encoding .. so, not always 3 bytes per char. did i misinterpret what you said on that wikitech-l thread? [18:35:48] jamie crespo clarified this with: "'utf8' in MySQL is 3-bytes UTF-8. "Real" UTF-8 is called in MySQL utf8mb4." .. so, never mind. :) [18:53:14] subbu: up to 3 bytes, yeah I phrased that poorly