[03:58:32] <{Soap}> Im curious to see the FOP banner people are talking about [03:58:45] <{Soap}> since it now apparently excludes the USA, I cant make it come back [03:59:11] <{Soap}> https://en.wikipedia.org/wiki/Wikipedia_talk:Freedom_of_Panorama_2015#Proposal:_Banner_alert_campaign <- is it this? [04:00:56] delete the related cookie in your browser and it will come back ;) [04:03:55] <{Soap}> no i dont think so [04:04:06] <{Soap}> they exclude all USA users. but its OK because i found them [04:04:08] <{Soap}> https://commons.wikimedia.org/wiki/Commons:Freedom_of_Panorama_2015/Proposed_messages#Vote [09:43:11] Hey Nikerabbit any progress regarding https://phabricator.wikimedia.org/T98004#1408483 [09:49:20] xzise: you don't follow gerrit emails? [09:49:25] it's at review https://gerrit.wikimedia.org/r/#/c/221610/ [09:50:28] to be honest not really :/ the emails aren't really good formatted to easily be read and then (at least I am) get so much spam which I don't want to know [09:50:58] maybe I haven't configured it properly [09:53:47] It sure takes a lot of configuration/magic to fix the brokenness of gerrit emails [09:53:52] Though Phabricator is probably worse [09:54:15] However the personal dashboard can be useful, when properly cleaned up :) [09:54:28] Hmm I'm not sure regarding Phabricator [09:54:54] Maybe I'm not subscribed to enough bugs but it's okay and there is not so much junk around it [10:02:56] xzise: +1 for sanity check is enough [10:04:10] Nikerabbit, is it actually possible to have a “blacklist” [10:04:52] like when there is "%$[^a-zA-Z0-9]*$." in the text? [10:12:32] xzise: I don't understand [10:13:59] Sometimes we had people translate the text in the brackets which won't work [10:14:13] now your change will help there as it will warn that there is a variable missing [10:14:29] but they still could add %(translated). [10:14:44] Although maybe it's not so importanted [10:14:48] *important [10:14:57] as it should be obvious now that it won't work [10:15:19] and it won't catch when someone uses a translation which is a-zA-Z0-9 [10:18:06] We just had a few which didn't use the place holders properly, e.g. by translating the name of the place holder or missing the character after the place holder. So I though we should maybe warn about those cases. But of course if it warns that the placeholder is missing (because the name was translated or the character after the brackets was missing) it should be sufficient. [10:24:41] xzise: it does catch unknown variables which match the pattern [10:26:00] Nikerabbit, stupid question but what should it "catch"? [10:26:08] invalid placeholders? [10:26:15] Or actual placeholders [10:26:22] which must be present in both languages [10:26:37] (I might not be the best guy for that as I don't really use TWN) [10:26:46] xzise: both, missing and extra ones (if they match the pattern) [10:26:55] Hmm [10:27:03] Okay [10:27:04] so %(foo)s would result in two warnings (missing, and extra) [10:27:16] %(pälä)s one (only missing) [10:27:39] well it won't match %(pälä)s [10:28:24] So it searches (using the regex) in the English text and checks if all are present in the translations (if not they will be marked as missing) [10:28:35] and any additional matched in the translation are marked as extra? [10:28:47] In that case it should match any letter inside the brackets [10:29:21] xzise: well, you have a point [10:29:30] but then again, the warning about missing parameters is imho enough [10:30:07] the warning for additional parameters is more useful to detech outdated translations and typos like $5 in MediaWiki rather than catching incorrectly translated ones [10:30:14] What qualifies a missing parameter then? [10:30:37] xzise: anything that matches the pattern in source which is not present in translation [10:30:42] Also Okay [10:30:45] -Also [10:30:54] but extras aren't allowed either [10:31:13] If the English is "%(foo)s" and the translation "%(foo)s %(bar)s" it will fail [10:31:30] nope [10:31:35] ? [10:31:46] umm, yes I mean [10:32:44] if the translation is "%(foo)s %(위키백과)s" then no warning is produced [10:32:50] but I find that very unlikely case [10:32:54] Because of the regex? [10:32:59] yep [10:33:09] and if the regex would be "%$\w+$[...]"? [10:33:12] it would warn [10:33:41] Or "%$[^)]$[…]" [10:33:50] ehr "%$[^)]+$[…]" [10:33:59] yes [10:34:27] Then why don't we use that regex? [10:34:37] though there are actually two regexps, one for the insertable and one for the checks, in the patch there is only the one for insertables [10:34:38] Let me check if Python allows unicode names [10:35:02] Okay names can use unicode [10:35:13] so %(위키백과)s would cause an error [10:35:20] because we don't provide 위키백과 [10:35:35] hi [10:35:36] is there a way to make central notice display a different banner/link for mobile? [10:35:47] we = pywikibot [10:35:53] https://meta.wikimedia.org/wiki/Freedom_of_Panorama_in_Europe_in_2015 is for desktop and https://meta.wikimedia.org/wiki/Freedom_of_Panorama_in_Europe_in_2015/mobile is for mobile [10:37:03] And if we can warn about extra parameters (especially if they produce incompatible translations) I'd prefer if it does that too ;) [10:39:11] updated that patch, but I need to make another patch for the checkers [10:39:55] Okay you might right that Unicode placeholders are unlikely as it'll warn about missing parameters [10:40:30] ToAruShiroiNeko: yes, you can make different banners and check the desktop/mobile checkbox in either [10:40:45] (With both banners assigned to the same CentralNotice campaign.) [10:42:10] xzise: please check the regex at https://gerrit.wikimedia.org/r/221610 [10:42:27] it was also accidentally made greedy, now it is again ungreedy [10:42:54] altough it doesn't matter [10:43:03] o.O? [10:45:35] Nemo_bis great! [10:45:48] can I actually make the edit myself and someone else can approve it? [10:46:09] I am a basic user with basic access [10:46:17] I am as basic as one can be [10:46:26] xzise: ? [10:46:35] Why doesn't greediness matter? [10:46:57] in "%(old)s was named to %(new)s" it will match all of it [10:47:00] won't it? [10:47:15] xzise: because ) is not allowed to match inside [10:47:21] not in this version nor in the previous one [10:47:23] ^^ [10:47:25] Ah right [10:47:38] so either +*? or [^)]+ [10:47:46] -.- [10:47:48] .+? [10:47:50] not +*? [10:48:29] +? and *? would be ungreedy [10:48:36] I know [10:48:42] that's why it's ".+?" [10:48:47] but because it has the /U flag which makes ungreedy the default, those are flipped [10:48:52] oh [10:48:57] I thought it was unicode? [10:49:02] that is u [10:49:08] and I make this mistake almost every time [10:49:11] (stuff, I actually don't understand what php is doing there) [10:49:17] okay [10:49:22] umm [10:49:32] I wonder whether I should actually add u there now [10:49:36] if + is ungreedy anyway [10:49:46] you could replace the character class by "." [10:50:32] hmm [10:50:34] yes I could [10:50:53] But it's up to you if you want to bother [10:51:01] the current one does the same [10:51:15] regex nitpicking, even better than Italians discussing soccer formations [10:51:28] I'm not sure about the u-flag though [10:52:04] Nemo_bis italians discuss soccer formations? [10:52:35] If they discuss formations it's football formations :P [11:01:55] exactly [11:02:02] :D [11:02:16] who can edit the central notice? [13:10:31] Git/Gerrit repos question: When I run "ssh -l aklapper -p 29418 gerrit.wikimedia.org gerrit ls-projects" I get the list of projects in Gerrit. Now wondering if I'd get exactly the same list of projects for Git or if we have Git projects that are outside of Gerrit (and if so, why). [13:10:33] Anyone knows? [13:12:15] repos on GithHub that are not on Gerrit! ;) [13:23:13] Glaisher, are there repos in GitHub mirrored to Wikimedia Git, instead of the other way round? [13:28:07] andre__: there are some repos in GitHub which are not mirrored in gerrit/git [13:28:55] andre__: biggest example is SMW [13:29:15] ok, let me rephrase my question to add "Wikimedia" everytime I wrote "Git" [13:29:40] When I run "ssh -l aklapper -p 29418 gerrit.wikimedia.org gerrit ls-projects" I get the list of projects in Wikimedia Gerrit. Now wondering if I'd get exactly the same list of projects for Wikimedia Git or if we have Wikimedia Git projects that are outside of Wikimedia Gerrit [13:31:44] andre__: is the context https://phabricator.wikimedia.org/T103292 ? [13:32:01] I think korma is only about projects hosted primarily in gerrit.wikimedia.org, correct? [13:32:23] Because Wikimedia has so many repos outside of gerrit that it would impossible to chase them all [13:34:10] Nemo_bis, yes, that's the context [13:34:27] The Code Review part of korma only targets Wikimedia Gerrit, indeed [13:34:37] however there's also a Code Management part of korma. not sure about that one myself either [15:23:27] uuh https://github.com/WPO-Foundation/webpagetest/issues/20#issuecomment-117684053 [15:27:33] SMalyshev: will be a bit late for meeting [17:04:05] Krinkle_: ugh, could you update https://en.wikipedia.org/wiki/MediaWiki_talk:Anonnotice with what the proper way to do logged-in only sitenotices is? [17:04:17] and thanks for the revert [18:54:32] legoktm: The proper way is to just leave Anonnotice empty [18:54:34] properly empty [18:54:57] The default value when the message isn't overridden is '-' which makes it pass through to sitenotice. [18:55:08] empty string will make it disabled for anons [19:30:47] The proper way is to avoid changing the proper way continuously. :) [19:31:16] relevant xkcd: https://xkcd.com/927/ [19:41:14] anomie: I mentioned the API query continue change on https://wikitech.wikimedia.org/wiki/Deployments and tweaked https://www.mediawiki.org/wiki/API:Query?type=revision&diff=1698208&oldid=1686296 [19:42:27] spagewmf: ok [19:46:49] anomie: how's the APImageddon looking? :) I see the Tech News announcement https://meta.wikimedia.org/wiki/Tech/News/2015/27 , I'd like to note it on https://www.mediawiki.org/#mainpage_newscell [19:47:54] spagewmf: Haven't heard anything from people so far [19:48:13] anomie: also, AIUI your bot list includes everyone not specifying a continue or rawcontinue. But if a bot didn't process continuation data before, nothing bad will happen [19:49:53] spagewmf: Correct [19:51:25] alternatively relevant xkcd: https://xkcd.com/1172/ [19:52:17] xzise: :) [20:41:42] anomie: Will the "Formatting of continuation data has changed. ..." warning ever go away? if you don't care about continuation data you can leave it off, but I'm not crazy about API examples returning warnings [20:42:11] I mean "leave off continue=" [20:43:06] spagewmf: Once REL1_26 is cut, probably. [20:47:12] anomie OK, I'll leave continue= off sample queries that don't deal with continuation data. (unlike formatversion=2 which we should add to sample queries). Thanks