[00:03:51] Oh boy [00:11:44] (03CR) 10Adamw: [C: 031] Add back serialize() calls for internal errors [extensions/Cite] - 10https://gerrit.wikimedia.org/r/86372 (owner: 10Bartosz Dziewoński) [02:13:23] (03PS1) 10Brian Wolff: Move getRotation to the base class [core] - 10https://gerrit.wikimedia.org/r/86404 [02:28:01] bugs in UploadWizard? never heard of 'em! [02:28:12] Nooope [02:28:35] YuviPanda|zzz: I successfully amended https://en.wikibooks.org/w/index.php?title=Cookbook:Penne_with_Hearty_Meat_Sauce&stable=0 with UploadWizard today, e.g. :) [02:29:41] ah [02:29:41] nice [02:30:00] I apparently went to sleep at 9:30PM and woke up at 8AM :| [02:30:12] marktraceur: did you uplaod them from your fierfox phone? :) [02:30:18] Noooope [02:30:25] YuviPanda: The camera on it is shite [02:30:31] I used the n900 as a digital camera [02:30:41] * YuviPanda considers making a 'just like the OS!' joke [02:30:50] * YuviPanda declines to make the joke [02:31:19] Boy I sure am glad YuviPanda didn't make a joke about FirefoxOS just now [02:31:25] That would have been really shitty of him [02:31:28] And pretty fucking unfair [02:31:31] yeah, me too [02:31:34] But luckily he's not a total assface [02:32:24] yeah, it's not like the OS is actually people clinging to the DOM everywhere... [02:33:50] * marktraceur batters YuviPanda with DOM nodes [02:34:06] very useful, yes :P [02:34:33] I'm doing my bit of good (or evil?) by taking my dad to the 10th anniversary of Tamil Wikipedia event [02:34:45] they've a workshop, and he actually loves the language (unlike me!) [02:35:18] Coolio [02:35:43] I was trying to get a friend to contribute to an incubator-hosted language [02:36:12] https://incubator.wikimedia.org/wiki/Wp/miq [02:37:18] I should implement the python + varnish shorturl proposal on the way [02:43:37] I tend to wake up around the 'dead' zone these days. sucks very badly. [02:43:44] where most people are just winding down [02:43:49] hurr durr [02:44:14] YuviPanda: But look! A bawolff! [02:44:32] Indeed [02:44:38] awww! [02:44:39] except I'm going to bed soon [02:44:44] Nuuuu [02:44:47] You can't go to bed [02:44:54] see! :P [02:44:57] proves my point, etc [02:45:04] YuviPanda: Fine, *I'll* stay up with you at least [02:45:12] aww marktraceur! [02:45:15] I went to bed at 3 this morning, I'm sure I'll be up a while [02:45:20] I want to make PagedTiffHandler work with VipsScaler before I go to bed [02:45:26] it's okay, I'm going to head into a bunch of tamil wikimedians in an hour [02:45:33] because its been annoying me for a while that it doesn't [02:45:40] YuviPanda: You might want to rephrase that [02:45:43] Or maybe you don't [02:45:44] I dunno [02:46:13] bawolff: re: 'will not caching the parsed campaign info cause problems?', an API request for 140 campaigns now takes about... 15s [02:46:15] on the cluster [02:46:36] bawolff: doesn't affect Campaign: pages as such (thank squid!) [02:46:47] yay Squid! [02:46:48] so looking for ways to cache it. [02:47:10] yeah, I saw that bug, the amount of time it took was quite a bit longer then I thought it would [02:47:12] looks like I'll have to track dependent template lists, insert them into templatelinks, and invalidate them myself. [02:47:22] the a mount of time it took to surface the bug? :P [02:47:45] More I don't expect 140 simple pages to take over 20 seconds [02:47:54] bawolff: yeah, me neither [02:48:08] bawolff: an average of 2 parses of mostly one short line of wikiext per campaign. [02:48:39] bawolff: we added profiling in cluster, 100-200ms for each parse call. Adds up a fair bit when asked for ~140 [02:49:59] and then handle per-language invalidation [02:50:37] sad. &forceprofile=true doesn't seem to work on api [02:53:46] YuviPanda: uclimit=19 goes in about 2-3 seconds, where uclimit=20 goes in 6-7 seconds. That doesn't quite seem a linear slow down [02:54:23] I found uclimit=50 go for about 6-7s, and ~140 at about 15-17 [02:54:26] felt linear? [02:54:34] bawolff: plus, some campaigns have almost no wikitext... [02:55:28] hmm, maybe that's it - but it did seem that wlm-de-ni added a whole 3 seconds (otoh this is not a rigorus test i just did) [02:57:04] bawolff: 4 wikitext parses [02:57:15] bawolff: plus a tutorial: true, although I'm not sure if that should affect anything [02:57:26] huh, and I just did it now and it took 37 seconds, so I don't know wtf is going on [02:57:54] bawolff: also note that the API by default calculates stats - so that also has to be factored in [02:58:02] bawolff: although we cache those... [02:58:08] bawolff: though not for long! [02:59:06] Hmm, I just did $ time wget -S 'https://commons.wikimedia.org/w/api.php?action=query&list=allcampaigns&ucenabledonly=&uclimit=19' [02:59:24] 4 times. It took 3.5, 3.5, 4, and 18 seconds [02:59:57] which is odd [03:00:05] they should be showing up on graphite [03:00:07] let me loo [03:00:07] k [03:00:50] bawolff: https://graphite.wikimedia.org/render?width=500&from=-12hours&until=now&height=380&target=UploadWizardCampaign.getParsedConfig.tavg&target=UploadWizardCampaign.parseValue.tavg&uniq=0.4038376593962312 [03:01:08] bawolff: you *should* be able to view that/ [03:01:09] ? [03:01:17] yeah I can [03:03:18] bawolff: getTotalContributionsCount averages 20ms [03:03:50] Hmm, so why is this sometimes super slow, and sometimes not [03:03:56] getUploadedMedia is under 2s [03:04:06] How long are you caching those queries. More then 15 seconds I assume? [03:04:07] err [03:04:07] 2ms [03:04:10] or 2ms [03:04:17] this doesn't say units [03:04:24] bawolff: yeah, about 10m [03:04:27] by default [03:05:19] I would assume the fact I've loaded the api so many times would make all those queries be cached [03:05:32] I would too [03:05:59] unless different one's out of the 20 are being hit when I load, if their expiry is staggered. But even still I wouldn't expect it to slow things down that much [03:06:14] me neither [03:06:43] I plan on reproing this locally and making sure i'm fixing the right thing before actually fixing it [03:07:50] I really wish that forceprofile=true worked on the api, since then it would be easy to see which part for a specific request was slow [03:12:49] As a totally unrelated aside, it might be cleaner if instead of storing campaign_enabled in a separate table, it was stored in the page_props table [03:25:04] oh ?forcetrace does work on the api, that's interesting [03:26:41] YuviPanda: ok I just did https://commons.wikimedia.org/w/api.php?action=query&list=allcampaigns&ucenabledonly=&uclimit=15&servedby=&format=xmlfm&uccontinue=20&forcetrace=true which took 11 seconds (vs normal of about 6 seconds) [03:27:15] on that request 2 seconds went to the db query [03:27:17] 2.638305 -3.3 < UploadWizardCampaign::getTotalContributorsCount [03:28:23] (and it did it a bunch of other times, most of those the db query took about 0.7 seconds) [03:28:59] bawolff: I've used the new gallery mode twice today, <3 [03:29:04] modes* [03:29:10] whoo! [03:29:29] YuviPanda: But the parsing part seems to also take enough time that it significantly slows things down [03:35:19] YuviPanda: Maybe preg_match( '/^

(.*)\n?<\/p>\n?/sU', $parsed, $m ) is slow [03:35:52] no, I actually don't think that's the case [03:37:41] but I still think that regex might not be speediest (With the U flag for ungreedy), and that there should be a wfProfileIn() around it just to be sure, but I don't think its a significant issue [03:42:50] bawolff: ah, hmm. ok [03:43:16] bawolff: it has been the same from the start tho - it is copy pasted from inside OutputPage [03:43:22] honestly, the more I look at it, the more I think I'm wrong, and the regex has nothing to do with it [03:43:59] if we average 150ms and have 140 campaigns, 150ms * 140 = 21s [03:45:10] bawolff, that regex is taken from OutputPage, I doubt it's so slow [03:45:28] bleh, I should read before writing [03:45:33] ;) [03:46:14] YuviPanda, any luck with LinksUpdate? [03:46:16] yeah, I got a little prematurely exicited [03:47:20] MaxSem: on the way to a tamil wiki meetup now, and was distracted by other things yesterday. Will look at it later tonight [03:57:38] YuviPanda: So yeah, I think my sudden increase in time i got before was just because I happened to hit when that query wasn't cached [03:57:51] sounds possible, yeah [04:07:36] (03CR) 10Parent5446: [C: 031] "Haven't tested." [core] - 10https://gerrit.wikimedia.org/r/65299 (owner: 10Mwalker) [04:09:07] (03CR) 10Parent5446: [C: 031] "Haven't tested." [core] - 10https://gerrit.wikimedia.org/r/65868 (owner: 10Mwalker) [04:14:02] hmm PagedTiffHandler was failing for me locally because it wanted to make an 800mb temporary file in the process of verifying that the file I'm uploading really is a tiff file.... [04:27:37] (03CR) 10Ori.livneh: "It was hasty, especially since the content of the README was only very recently collaboratively edited on MW.org: https://www.mediawiki.or" [core] - 10https://gerrit.wikimedia.org/r/86353 (owner: 10Yuvipanda) [04:29:17] (03PS2) 10Brian Wolff: Move getRotation and getImageArea to the base class [core] - 10https://gerrit.wikimedia.org/r/86404 [04:30:18] (03CR) 10jenkins-bot: [V: 04-1] Move getRotation and getImageArea to the base class [core] - 10https://gerrit.wikimedia.org/r/86404 (owner: 10Brian Wolff) [04:30:33] hisss [04:31:17] (03PS3) 10Brian Wolff: Move getRotation and getImageArea to the base class [core] - 10https://gerrit.wikimedia.org/r/86404 [04:31:20] guess I should run git show before hitting the button [04:32:03] bawolff: Yuup [04:36:07] (03PS1) 10Brian Wolff: Make PagedTiffHandler work with VipsScaler. [extensions/PagedTiffHandler] - 10https://gerrit.wikimedia.org/r/86413 [05:04:37] (03CR) 10Ori.livneh: [C: 032] "Excellent, thanks!" [vagrant] - 10https://gerrit.wikimedia.org/r/86336 (owner: 10Yurik) [05:04:41] (03Merged) 10jenkins-bot: Fix settings and documentation for remote_debug role [vagrant] - 10https://gerrit.wikimedia.org/r/86336 (owner: 10Yurik) [05:09:58] (03CR) 10Ori.livneh: [C: 04-1] "(1 comment)" [core] - 10https://gerrit.wikimedia.org/r/86361 (owner: 10MaxSem) [05:11:35] (03PS2) 10Brian Wolff: Make PagedTiffHandler work with VipsScaler. [extensions/PagedTiffHandler] - 10https://gerrit.wikimedia.org/r/86413 [05:15:22] (03Abandoned) 10Ori.livneh: Use lessc::addParsedFile instead of custom attribute [core] - 10https://gerrit.wikimedia.org/r/86360 (owner: 10Ori.livneh) [05:16:02] (03PS2) 10Ori.livneh: less: Use new addParsedFile method, replacing embeddedFiles hack [core] - 10https://gerrit.wikimedia.org/r/86249 (owner: 10Krinkle) [05:16:07] (03CR) 10Ori.livneh: [C: 032] less: Use new addParsedFile method, replacing embeddedFiles hack [core] - 10https://gerrit.wikimedia.org/r/86249 (owner: 10Krinkle) [05:20:31] (03Merged) 10jenkins-bot: less: Use new addParsedFile method, replacing embeddedFiles hack [core] - 10https://gerrit.wikimedia.org/r/86249 (owner: 10Krinkle) [05:28:17] (03PS27) 10Physikerwelt: Math 2.0 [extensions/Math] - 10https://gerrit.wikimedia.org/r/85801 [05:31:30] (03CR) 10Brian Wolff: [C: 04-1] "This change depends on a small change to VipsScaler I have yet to commit (basically just adding : to the end of the file name)" [extensions/PagedTiffHandler] - 10https://gerrit.wikimedia.org/r/86413 (owner: 10Brian Wolff) [05:34:17] (03CR) 10Brian Wolff: "Actually segfault seems to only happen when specifying non-integer shrink sizes" [extensions/PagedTiffHandler] - 10https://gerrit.wikimedia.org/r/86413 (owner: 10Brian Wolff) [05:41:15] (03PS2) 10MZMcBride: Correct attribution for Wikimedia project visitors and restore grammar [core] - 10https://gerrit.wikimedia.org/r/86394 (owner: 10Adamw) [05:47:30] (03CR) 10MZMcBride: "Time is precious; life is short." [core] - 10https://gerrit.wikimedia.org/r/86394 (owner: 10Adamw) [06:23:21] (03CR) 10Ori.livneh: "What is 'ConfirmAndMigrateUser'?" [extensions/CentralAuth] - 10https://gerrit.wikimedia.org/r/64216 (owner: 10Pgehres) [06:30:36] (03Abandoned) 10Hashar: Jenkins job validation (DO NOT SUBMIT) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/86118 (owner: 10Hashar) [06:32:09] (03PS1) 10Brian Wolff: [WIP] Add support for page numbers to VipsScaler. [extensions/VipsScaler] - 10https://gerrit.wikimedia.org/r/86416 [06:34:21] (03PS3) 10Brian Wolff: [WIP] Make PagedTiffHandler work with VipsScaler. [extensions/PagedTiffHandler] - 10https://gerrit.wikimedia.org/r/86413 [07:11:18] (03PS1) 10TTO: Standardise wording of verbs relating to revision deletion [core] - 10https://gerrit.wikimedia.org/r/86419 [07:11:58] bd808|BUFFER: just a fyi, the tawiki community (i'm at their meetup) are incredibly happy with matjax, can't wait for it to be enabled [07:12:13] mostly because texvc dies if you give it something not ASCII [07:19:22] (03PS1) 10Physikerwelt: Add local LaTeXML webservice for the Math role [vagrant] - 10https://gerrit.wikimedia.org/r/86420 [07:31:46] (03CR) 10Rillke: "jQuery is the framework MediaWiki has chosen. I do, however agree that an anonymous function might be more intuitive for people who are no" [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/85324 (owner: 10Rillke) [07:34:43] (03CR) 10MarkTraceur: "Well, I mean...JavaScript is the language we've chosen. jQuery is just a library we use for some things." [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/85324 (owner: 10Rillke) [07:43:16] (03PS8) 10Rillke: Replacing Function.prototype.bind with jQuery.proxy [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/85324 [07:44:06] (03CR) 10Rillke: "Still +5 lines :)" [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/85324 (owner: 10Rillke) [07:45:28] Urgh. Ah well. [07:47:27] (03PS9) 10Rillke: Replacing Function.prototype.bind with jQuery.proxy [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/85324 [08:00:20] (03PS2) 10Physikerwelt: Add local LaTeXML webservice for the Math role [vagrant] - 10https://gerrit.wikimedia.org/r/86420 [08:04:14] (03CR) 10Physikerwelt: "For some reason, I can not test it on windows. And I did not mange to test it on a labs instance either." [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [08:43:20] (03CR) 10Ori.livneh: "What error(s) are you seeing?" [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [08:57:34] (03CR) 10Physikerwelt: "I don't see any output, if I try to connect to the vagrant instance via ssh" [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [08:59:12] (03CR) 10Ori.livneh: "Sorry, I meant on Windows, using the local Vagrant." [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [09:03:41] (03CR) 10Physikerwelt: "First, after solving the issues with the filesystem. I get" [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [09:57:48] (03PS3) 10Yuvipanda: Add local LaTeXML webservice for the Math role [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [10:05:00] (03CR) 10Yuvipanda: "Patchsets 3 fixes errors I got on my local instance (OS X host). Couldn't test it fully since apparently I'm on a connection slower than E" [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [10:13:08] (03PS1) 10Nemo bis: Complete rename to EmailableUser [extensions/CentralAuth] - 10https://gerrit.wikimedia.org/r/86421 [10:53:12] (03PS4) 10Yuvipanda: Add local LaTeXML webservice for the Math role [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [10:53:38] (03CR) 10Yuvipanda: "PS4 is just whitespace fixes" [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [11:27:42] (03CR) 10Bartosz Dziewoński: [C: 032] Add back serialize() calls for internal errors [extensions/Cite] - 10https://gerrit.wikimedia.org/r/86372 (owner: 10Bartosz Dziewoński) [11:27:50] (03Merged) 10jenkins-bot: Add back serialize() calls for internal errors [extensions/Cite] - 10https://gerrit.wikimedia.org/r/86372 (owner: 10Bartosz Dziewoński) [12:31:27] (03CR) 10Hashar: "(2 comments)" [tools/release] - 10https://gerrit.wikimedia.org/r/86259 (owner: 10Hashar) [12:31:37] (03PS3) 10Hashar: switch methods to be context (options) based [tools/release] - 10https://gerrit.wikimedia.org/r/86259 [13:10:32] Hello, I understand that running arbitrary lua code on page is bad. but why also preventing running builtin's functions (https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual). For example, if you just need to run os.date("%Z"), you'll need to create a wrapper function. [13:35:42] (03Abandoned) 10Hashar: (bug 34750) Accessibility improvement for embedded images [core] - 10https://gerrit.wikimedia.org/r/9967 (owner: 10Kai Nissen (WMDE)) [13:42:03] (03Abandoned) 10Hashar: (bug 34778) Extension Memento: Initial Submit [extensions/Memento] - 10https://gerrit.wikimedia.org/r/29812 (owner: 10Hariharshankar) [13:42:08] (03Abandoned) 10Hashar: (bug 34778) Extension Memento: Improvements after previous review. [extensions/Memento] - 10https://gerrit.wikimedia.org/r/32237 (owner: 10Hariharshankar) [13:42:11] (03Abandoned) 10Hashar: (bug 34778) Extension Memento: Improvements after review. [extensions/Memento] - 10https://gerrit.wikimedia.org/r/32238 (owner: 10Hariharshankar) [13:42:16] (03Abandoned) 10Hashar: (bug 34778) Extension Memento: Improvements after review [extensions/Memento] - 10https://gerrit.wikimedia.org/r/32239 (owner: 10Hariharshankar) [13:47:01] (03Abandoned) 10Hashar: Adding the files [extensions/MobileSections] - 10https://gerrit.wikimedia.org/r/36366 (owner: 10Jnfsmile) [13:47:11] (03Abandoned) 10Hashar: Fixes mentioned in review results [extensions/MobileSections] - 10https://gerrit.wikimedia.org/r/36646 (owner: 10Jnfsmile) [13:52:12] (03PS4) 10Hashar: Add Basic Tamil Grammar rules, unit test [core] - 10https://gerrit.wikimedia.org/r/32855 (owner: 10Logicwiki) [13:53:01] (03CR) 10jenkins-bot: [V: 04-1] Add Basic Tamil Grammar rules, unit test [core] - 10https://gerrit.wikimedia.org/r/32855 (owner: 10Logicwiki) [13:54:22] (03CR) 10Brion VIBBER: "(2 comments)" [core] - 10https://gerrit.wikimedia.org/r/86361 (owner: 10MaxSem) [14:51:18] (03CR) 10Brion VIBBER: "Mark, are you going to do another version of this patchset or should we merge it and let you address the concerns in a subsequent commit?" [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/78341 (owner: 10MarkTraceur) [14:53:13] anyone? [14:53:42] ytrezq: ? [14:53:54] his question is about 2h ago [14:54:13] * brion reads backscroll [14:54:22] ytrezq: what do you mean "builtin functions"? even basic commands can be quite dangerous [14:54:30] ytrezq: so in general you don't know what's going to be safe, so you start with no functions :) [14:54:57] an "os" module sounds dangerous in particular [14:55:36] http://www.lua.org/manual/5.2/manual.html#pdf-os.execute <- like this would be something you def wouldn't want on by default :) [14:55:36] Well, I don't how you can make something dangerous with this: https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual#Math_library [14:56:32] so if i'm reading that page correctly, the math library is available. [14:56:34] so…. ? [14:56:39] not sure i understand the question [14:57:40] The math library is availble but if you want to execute one of it's function directly, you need to create a module. [14:57:56] With a wrapper function inside [14:58:16] as i understand you need a module to run any lua code, that' show lua code gets run [14:58:19] or am i mistaken? [14:59:49] math library is a part of what i call � lua builtins functions �. I don't understand the fact you can't call them [15:00:42] Or moreover, the fact you'll need a workaround to do it (like creating a wrapper function). [15:01:28] sorry � the fact � I wanted to say: the reason [15:02:12] I still don't understand why you need to create a wrapper. Could you give an example of what you would want to do, and what you have to do now? [15:02:32] well my understanding is that you cannot call any lua functions directly at all; you write lua code such as functions in a module, then you invoke that module from a wiki page [15:03:51] i think what ytrezq is asking is 'why aren't "math", "os", etc available as modules to #invoke"? [15:03:59] in which case the answer is easy for "os" -- it's unsafe [15:04:10] for "math" -- i suspect it could be done [15:04:51] but in general i'd recommend against exposing low-level interfaces that way [15:05:06] a module should, in theory, be reimplementable (like if we switch from lua to something else in the future) [15:05:13] and it's nice to not tie lua-specific calls into all your wiki articles [15:05:31] as in just calling {{#invoke|math.exp(15)}} ? [15:05:56] hmm [15:06:03] brion: You're right I can't see calling a lua function (or a least math) directly is forbidden (the first question was to ask the reason since it is safe for some of them). And I can't see os used with os.date() is dangerous [15:06:08] also, looks like there's a specific parameter syntax [15:06:15] so someone would still have to write the module wrappers [15:06:19] just they could be bundled for convenience :) [15:07:00] it's probably fairly straightforward to write a generic wrapper function that calls a buildin module/table by name [15:07:09] MartijnH: It didn't worked: it look for a module name math.exp(15) [15:07:15] named [15:07:20] ytrezq: i would recommend using mw.language:formatDate instead of os.date though, since it'll tie in with the wiki's timezone and language settings [15:07:33] ytrezq, I wasn't implying that works, just asking if that is what you were looking for [15:08:13] brion: still the same problem: you can't call this directly [15:09:24] but just so that I'm sure I'm understanding what you're asking, you would like to be able to call lua "in line" from your {{#invoke}} call right? [15:09:59] I wanted to replace this page(https://fr.wikipedia.org/w/index.php?title=Wikip%C3%A9dia:Le_Bistro/Heure_d%27%C3%A9t%C3%A9&action=edit) with os.date("%Z") [15:11:38] MartijnH: I want to call a lua builtin's functions, in the same way from user's one in a module. You don't need to create a wrapper for calling code in amodule :D [15:12:35] ytrezq, right now, you can't call lua from anything else than a module (that is, something that is in a Module: namespace) [15:14:14] so what you would like, to call a "standard" library from a template is impossible. I'm thinking however if I could make a forwarder that does just that [15:15:44] apart from that, you will (probably) never be able to call anything from the os buildin, because of security reasons [15:15:50] I don't want to run arbitrary lua code. I am thinking declaring a function (or using thing like return) in an article should be forbidden. What i am asking is the reason for not running lua functions described here(https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual) since I don't really in what math.* are dangerous [15:16:12] would the syntax {{#invoke:dyn|math.exp|15}} work for you? [15:16:58] a module dyn that takes the first argument, parses it to elements, and passes the rest of the functions should be feasible, though I'm not sure if it is a good idea to go that route [15:17:48] MartijnH: that might be tricky with parameter types; does Lua do auto-coercion from strings to numbers etc? [15:17:57] No [15:17:59] it might require an explicit wrapper than knows the parameters better [15:18:16] scribuntu auto-coerces everything to string, so yeah, that may be tricky indeed [15:19:34] hrm, just when i think i'm clever... [15:20:22] brion: Have you seen how much funtion there is contents summary of that page: https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual [15:21:19] it could work with {{#invoke:dyn|math.exp|15|number}} but at that point you're so close to making a domain specific language that is so cumbersome you might prefer not to use it at all [15:21:38] ytrezq: so long story short: you're asking for something that's not consistent with the design of the scripting system. if you write the wrapper modules, you can use them. if you don't, too bad. [15:23:04] brion: Do you imagine the work of creating a wrapper module imply? I don't see the difference between disallow using them if it is allowing them running via a module if the code for doing already exist [15:24:35] it's not that they're disallowed. it's more like "if you want some milk, you have to open the refrigerator first and *then* reach in and get the milk" [15:25:01] By telling this, I have an idea of doing this simpler. But there is maybe a problem in mw.lua. https://fr.wikipedia.org/wiki/Module:Wrapper [15:25:05] the refrigerator *could* have a special milk dispenser, but it seems like it would be weird. [15:26:17] For me it is steel trying to do the work someone did already [15:28:11] you do know about parser functions in wikitext like {{#time}} right? [15:28:49] i don't think you need to be calling into lua for this in the first place [15:28:58] and if you did, you should really wrap it in a nice encapsulation [15:29:13] which is to say…. a module [15:29:36] Yes I have already used them. but for getting a string telling CEST/CET according the date, lua start to be better [15:30:14] because this page looks like a big hack: https://fr.wikipedia.org/w/index.php?title=Wikip%C3%A9dia:Le_Bistro/Heure_d%27%C3%A9t%C3%A9&action=edit [15:30:20] sounds like something i'd want to wrap in a module called timeZoneForDate or something [15:30:42] i can't imagine a reason to call os.date directly from a wiki page for this [15:31:17] then the module can be reused in many places [15:31:23] instead of manually copied from one to the other [15:31:41] there's no real equivalent with %Z parameter than os.date [15:34:09] after all I won't be talking of this. If i know what is the error in my code: https://fr.wikipedia.org/wiki/Module:Wrapper [15:35:12] well, for one, you're trying to call os, which if I'm not mistaken (and do correct me if I'm wrong brion) isn't available in scribunto [15:35:30] ytrezq: so that won't work for several reasons -- first the os module isn't available at all [15:35:44] second, the functions on the os module wouldn't be expecting mediawiki's frame object as a parameter [15:36:06] MartijnH: No you can use os functions inside a module [15:36:21] See: https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual#Operating_system_library [15:36:39] ah, looks like a couple functions do get re-exported. handy [15:36:47] Of couse the unsafe ones have been removed [15:36:49] then you can just call that from your own module…. i don't see the problem here :) [15:37:16] Me two: I don't understand the error [15:37:29] oh the error you're getting is because of the things i said already [15:37:35] no surprise there [15:37:38] (Erreur de script) [15:37:51] just…. don't try to do what you're doing, it's not going to work. [15:38:03] call the function from a module, and invoke the module from your template. [15:43:39] again: there is more than 200 functions to wrap? [15:43:43] again: there is more than 200 functions to wrap! [15:44:33] ytrezq, have you made any modules yet? [15:46:39] I prefer searching rather than going strait ahead and seeing they were a very shorter way to do when the work is finished. :) [15:47:14] It is also a kind of green computing :D [15:47:59] ytrezq, the {{#invoke}} syntax passes to the module what is known as a parser frame. You will always have to get the paramters from the parser frame, and possibly parse them to numbers, and then pass them on to "normal" lua functions [15:48:16] thus autowrapping is impossible [15:49:08] thus calling anything from lua standard libraries directly from any {{#invoke}} call is impossible [15:49:58] https://fr.wikipedia.org/wiki/Module:Wrapper isn't trying to call lua standard libraries directly [15:50:25] no, it is trying to pass the os table back, and you're trying to call functions on that table, which won't work [15:50:46] because you'll be passing in a parser frame, and functions in the os table will not be expecting a parser frame [15:51:50] I would really advice you first to play around with it a little, see how it is supposed to be used [15:53:35] * YuviPanda pokes brion [15:53:40] brion: been using vagrant? :D [15:54:04] YuviPanda: yes, though i wish that virtual box and parallels didn't KILL EACH OTHER on my mac if i leave them both on [15:54:11] hehe :P [15:54:21] gotta remember to do 'vagrant down' before firing up windows vm [15:54:40] brion: there exists https://github.com/yshahin/vagrant-parallels, fwiw [15:54:52] yeah i'll have to check that out [15:54:59] i suspect it'll take some work to make work [15:55:24] brion: it's pretty active [15:55:34] then maybe not too much work :D [16:03:07] brion, now that you're distracted anyway, if you could find the time and willingness I would love to see more comments from staffers on https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(proposals)#Wikimedia_Foundation_employees_are_members_of_the_community [16:03:21] though I can easily understand if you wouldn't [16:03:57] ugh, pages of crap to wade through [16:04:00] i'll probably pass :) [16:04:39] yeah, we're good at creating pages of crap over at en.wp D: [16:11:11] that's a lot of people saying the same things over and over again [16:12:45] still, I like any place that dares implying that en.wp and the WMF aren't enemies, and have the same goals. [16:12:59] maybe it takes a lot of repetition to get that in to peoples thick skulls [16:13:20] heh, indeed. [16:13:45] MartijnH: I personally think that I should get involved in conversations where me saying things has some amount of impact. I don't see anything there I can say that others haven't said more eloquently, so... [16:14:56] YuviPanda, every staffer engaging into dialog with any en.wp editor as an equal is a happy step forward IMO. Even if the same thing is being said, it might help the feeling that we're trying to work together to get forward [16:16:15] i just don't find discussions that are mostly people whinging to be useful to anybody [16:16:20] yeah, +1 [16:16:23] i prefer to get involved in discussions that involve getting stuff done [16:16:31] I agree with what you said MartijnH, but not when framed as how it is now [16:16:36] even if it would be saying the same, or being idle chit chat about their favourite coffee. The thing that is hurting us most atm is the lack of perceiving the other as a person trying to do their best for the mutual goal rather than some remote party or entity (let alone adversary) [16:16:57] I used to idle chit chat on -en, fwiw. [16:17:05] true [16:18:51] I probably should. I think I quit during one of the periods of too much vitriol [16:19:27] I don't blame you [16:20:09] of course one general problem is that the definition of "the community" is frequently taken as "a subset of Wikimedia contributors and users who talk loudly at each other" [16:20:48] meanwhile the vast majority of people who use our sites apparently are only vaguely aware that editing exists [16:21:17] so i find it tough to make generalizations [16:24:25] I don't really think that we should include our readers in our definition of the community though. What I regard as the community is the people who help create the projects. If we through in the readers in that group, the group becomes so wide that it is impossible to say anything useful about the community anymore, or to act in a way that affects the community [16:25:09] that leaves the vast group of people who just edit stuff and create content in the middle obviously, I'm not trying to marginalise them [16:26:13] well i'd rather marginalize a few of the editors or developers than all 7 billion potential readers, in many cases [16:26:26] (03CR) 10Siebrand: [C: 032] Update README [core] - 10https://gerrit.wikimedia.org/r/86352 (owner: 10Yuvipanda) [16:27:04] and i'd rather marginalize "current editors who are loud about their workflows" than every potential future editor [16:27:22] and that's a statement that's never going to go over well with those loud people. [16:27:45] I could see that (though I disagree, but that's another story altogether), I'm just saying that those readers, while important, should not be taken in to account when talking about the community [16:28:09] they're a different group, and have (obviously) different needs [16:28:13] if we're not considering our readers, why the fuck are we here [16:28:30] if this has become a project that exists just to exist, then we should pull the plug [16:28:30] that's not what I'm trying to say at all [16:29:04] sure [16:29:16] but some do seem to :) [16:29:20] what I'm trying to say is that we shouldn't take the readers along as a group when discussing the community. They are obviously an immensly important part of the project [16:29:20] {{citation needed} [16:29:26] (03Merged) 10jenkins-bot: Update README [core] - 10https://gerrit.wikimedia.org/r/86352 (owner: 10Yuvipanda) [16:29:49] (03CR) 10Siebrand: [C: 04-1] "(1 comment)" [core] - 10https://gerrit.wikimedia.org/r/86350 (owner: 10Yuvipanda) [16:32:17] MartijnH: I tend to disagree, in part because the whole system is meant to be participatory and inviting [16:32:43] people should be able to come in and go from "doing editing work" without being cast into a role as "other" [16:33:46] people can also have wildly different amounts of participation in different areas of the "community" [16:33:50] that's a good point, but it is a different role in the project. The transition between the roles would ideally be extremely simple, but it's still a different role [16:34:31] I use wikipedia as a consumer, and as a wikipedia consumer, I don't want to be bothered with editing issues, I just want to be presented with the best encyclopedia possible [16:35:25] at the same time, when I try to make it a better encyclopedia by editing, I want to have the best tools available to me for editing it, without those tools being nessecarily useful for me (or someone else) as a consumer [16:37:27] of course we have an organizational goal to attract more editors, make it easier for people to participate, add more gradations of participation... [16:37:35] …so i'm not convinced it's a binary "in/out of community" [16:38:14] that's interesting [16:39:01] I always did think of it that way. My anectotical evidence is that I only see people who get drawn in, and then become entrenched, but that's obviously from a biased sample [16:39:09] heh [16:40:10] the occational editor seems like a very difficult subject to study, though it would be very interesting [16:40:28] yes :) [16:40:42] they exist; by definition they're hard to talk to because they're not participating in talk pages and mailing lists [16:40:48] exactly [16:40:49] some don't even have accounts [16:41:05] and they're probably not really bothered with questionaires either [16:41:12] yep [16:41:33] i leave the complication of figuring out how to study those people to others :D [16:41:55] * brion just tinkers with software and pretends not to notice these are hard issues ;) [16:42:23] do you happen to know the ratio semi-static ip's vs fully dynamic ones (to draw an arbitrary line, say, changes less than once every three months to more than that)? [16:42:33] no idea unfortunately [16:42:35] lies brion [16:42:41] hehe [16:44:48] would someone like Eric Zachte maybe know more about this stuff? [16:45:07] he might… i dunno if we have good stats for IPness though [16:46:27] from my gut feeling I think that if we want to know more about the occational editor, the non-account group is a very important group, and ignoring that group because the data is hard to come by could make any research completely worthless [16:47:04] (not to mention the group that occationally signs in to their account) [16:49:28] *nod* [16:50:47] if you'd have to make an educated, or even wild guess, which percentage of constructive edits would you say comes from the 'halfway' group [16:52:40] brion: Sorry again I can't wrap any function at all: https://fr.wikipedia.org/wiki/Module:Wrapper . The simplest thing don't work [16:53:56] (03CR) 10Adamw: "(1 comment)" [tools/release] - 10https://gerrit.wikimedia.org/r/86259 (owner: 10Hashar) [16:54:05] (03CR) 10Adamw: [C: 032] switch methods to be context (options) based [tools/release] - 10https://gerrit.wikimedia.org/r/86259 (owner: 10Hashar) [16:54:07] ytrezq, how are you calling it? [16:54:09] (03Merged) 10jenkins-bot: switch methods to be context (options) based [tools/release] - 10https://gerrit.wikimedia.org/r/86259 (owner: 10Hashar) [16:54:30] ytrezq, I don't think that the parameter that you think is being passed in is getting passed in [16:54:45] MartijnH: it is written on the page [16:55:01] at the bottom [16:55:46] what you are passing in is a frame object of which the parameter arg is a table, and that table has a first parameter with the value " %Z" [16:56:21] what you seem to think is that you are passing in the string " %Z" [17:01:39] (03CR) 10Adamw: "Interesting, @Ori.livneh: thank you for posting that link. Now I see your detailed commit message about this, from July. It would be hel" [core] - 10https://gerrit.wikimedia.org/r/86353 (owner: 10Yuvipanda) [17:08:18] (03CR) 10Physikerwelt: "Working at the MediaWiki platform is so amazing." [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [17:08:53] (03CR) 10Adamw: "(2 comments)" [core] - 10https://gerrit.wikimedia.org/r/86394 (owner: 10Adamw) [17:10:19] (03CR) 10Yuvipanda: ":D" [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [17:12:00] MartijnH: emm: Sorry could you correct the error? [17:12:36] I definittly can't see what I do [17:12:38] please [17:13:20] well, the short is "it doesn't work that way" [17:13:28] I can maybe see if I can help you in a bit [17:13:36] right now I'm busy with something else [17:15:37] (03CR) 10Yuvipanda: "With that particular patch on, I get the following error:" [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [17:19:01] what you seem to be trying to do is function wrap.date(param) [17:19:01] return os.date(param.args[1]) [17:19:01] end [17:19:04] (03CR) 10MarkTraceur: "Probably the latter - addressing them here would mean a lot of rebasing down the line." [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/78341 (owner: 10MarkTraceur) [17:19:51] the param argument is the frame object [17:20:06] let me find you the documentation for the frame table [17:21:02] i've made the edit of what I think you're trying to do [17:23:40] https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual#Frame_object explains mere [17:23:48] s/mere/more/ [17:26:03] well, that's probably not what you're trying to do either [17:26:19] (03CR) 10Yuvipanda: "(and to test it before the patch gets merged - just clone mediawiki/extensions/Math yourself from the host machine, and then do 'git revie" [vagrant] - 10https://gerrit.wikimedia.org/r/86420 (owner: 10Physikerwelt) [17:30:17] (03CR) 10Brion VIBBER: [C: 032 V: 032] "Ok let's merge and fix up the later commits then :D" [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/78341 (owner: 10MarkTraceur) [17:30:18] (03Merged) 10jenkins-bot: Play with the information sent back from the hook [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/85367 (owner: 10MarkTraceur) [17:30:50] yay vagrant [17:31:03] wheee vagrant :D [17:31:06] i'll probably actually do some visual editor patches now that i don't have to worry about setting up parsoid manually [17:31:10] brion: +1 [17:31:34] brion: also once that patch from physikerwelt that I'm helping with gets merged, we can get people to hack on the math stuff too! [17:31:57] brion: btw, right now, there can be no non-ascii unicode inside a $tag, apparently :(
[17:31:57] (03CR) 10Brion VIBBER: [C: 032 V: 032] Add multilightbox library import/export script [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/82555 (owner: 10MarkTraceur)
[17:32:24] YuviPanda: yeah istr weirdness with all that
[17:32:32] istr?
[17:32:33] will have to catch up on the new stuff :D
[17:32:37] hehe :D
[17:32:41] i seem to remember/
[17:32:46] brion: I was at the Tamil Wikipedia 10th anniversary thing earlier today
[17:33:15] brion: one of the hardcore editors of the math articles there is a 60 year old retired ex-primary school math teacher, and it was awesome to see how excited she got when I showed her MathJax
[17:33:27] brion: since it actually lets her put things inside equations in Tamil
[17:33:27] :)
[17:33:30] oh cool!
[17:33:47] brion: they're planning on getting community consensus to enable it by default on tawiki
[17:33:59] sweeeeet
[17:34:12] brion: since the options there right now are 'no equations with tamil chars' or 'slightly slower equations'
[17:34:30] hopefully *that* doesn't meet with opposition from anywhere.
[17:34:41] heh
[17:35:22] brion: did you see the work with webfonts and mobile? :)
[17:37:32] i think i fixed the handling of comments in phpless
[17:37:52] not yet… that'll be fun :D
[17:38:00] it took me about 5 hours and resulted in a three-line diff
[17:38:06] heh
[17:38:45] ori-l: did you sleep at all?
[17:39:37] my son attached a clothespin to my ear at 8 am
[17:39:46] * brion is distracted installing XBMC on a Google TV
[17:39:58] ow
[17:40:08] it was kinda funny actually :P
[17:40:31] * YuviPanda googles clothespin
[17:40:36] oh
[17:40:36] that
[17:40:39] hehe
[17:40:52] hmm, I've always heard of that as 'clip'
[17:41:45] hmm, I don't think I can help the math patch at this point until I get a response - Reading makefiles doesn't seem to be my thing
[17:41:59] ori-l: also, how am I to set a require on a git clone?
[17:42:04] should i take a look?
[17:42:16] what do you mean?
[17:42:22] ori-l: looking at the code, I'd expect Exec['git clone {actual-title'] to work
[17:42:42] if not git::clone['{actual-title}'] even (not sure tho)
[17:43:08] ori-l: in https://gerrit.wikimedia.org/r/#/c/86420/
[17:43:22] one of puppet's many charming idiosyncrasies is that you use different letter case to declare resources and refer to declared resources
[17:43:26] right
[17:43:31] Git::Clone didn't work either
[17:43:34] so you'd do require => Git::Clone['actual-title'],
[17:43:35] I had to resort to using a before
[17:43:42] hm
[17:43:47] line 499, roles.pp on that patc
[17:43:48] h
[17:44:26] ori-l: also, the current setup seems a bit icky to me, in respect to packages. I now have to sortof repeat them three times - packages.pp, then the include, then the require
[17:45:04] I haven't read that patch yet
[17:45:06] let me read it now
[17:45:06] right
[17:45:08] ok
[17:45:27] wait what, it works now?!
[17:46:38]$