[00:00:23] New patchset: Reedy; "Add separate fa_sha1 field to filearchive table" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/17512 [00:00:52] New review: Reedy; "Needs rebasing" [mediawiki/core] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/4549 [00:00:54] csteipp, yep [00:01:26] New review: Reedy; "Needs rebasing too" [mediawiki/core] (master); V: 0 C: 1; - https://gerrit.wikimedia.org/r/17912 [00:01:28] Alright, I'm desperate. I'm going to put "french wikisource" scribunto into Google to see if Google knows where I read it. [00:01:31] New patchset: Reedy; "Show a "(blocked)" hint on Special:ListUsers/ActiveUsers" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/17945 [00:01:32] if anyone's interested, I finished work on a basic profiler yesterday, and now I'm adding some time and date functions to the default environment [00:01:35] Natterer: I am still awaiting your response to my earlier question, btw :) [00:01:49] What question? [00:01:55] Natterer: so, just to ensure I do not inappropriately speak down to you or speak in incomprehensible jargon: are you by any chance a software developer? [00:02:15] Natterer: (you may have also missed earlier things I've said to you & questions I've asked, I think) [00:03:01] csteipp, uh no hang on. That's with my 'deleted' -> 'rev_deleted' hack on. It's returning false. [00:03:12] sorry [00:03:17] Change merged: Reedy; [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/22198 [00:03:45] Krenair: ah, that makes more sense... although AaronSchulz may need to comment on why that's done that way. [00:04:09] I suspect it's because of getting archive rows [00:04:11] I'm guessing that's legacy, and we pass most things in with an object [00:04:18] see newFromArchiveRow [00:04:47] Yeah, that's probably it [00:04:51] 03(NEW) Enable extension Scribunto on en.WT - 10https://bugzilla.wikimedia.org/40031 normal; Wikimedia: Extension setup; (Amgine.Saewyc) [00:05:26] I've just found the s:fr: Scriptorum discussion. That wasn't it. Where *did* I read that? [00:06:01] 03(mod) Enable extension Scribunto on en.WT - 10https://bugzilla.wikimedia.org/40031 +comment (10Sumana Harihareswara) [00:06:49] New review: Krinkle; "The class clearing is not a problem if we don't use a class at all. The state before this change was..." [mediawiki/extensions/VisualEditor] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/22701 [00:07:13] New patchset: Mwalker; "Initial commit of the Unsubscribe extension!" [wikimedia/fundraising/DonationEmailUnsubscribe] (master) - https://gerrit.wikimedia.org/r/22679 [00:08:10] I've just found that page on the frame from a Lua person that TimStarling should read. But that's not it. [00:08:15] 03(mod) Enable extension Scribunto on en.WT - 10https://bugzilla.wikimedia.org/40031 +comment (10Amgine) [00:08:40] 03(mod) Enable extension Scribunto on en.WT - 10https://bugzilla.wikimedia.org/40031 normal->15enhancement (10Sam Reed (reedy)) [00:09:48] And I've just rediscovered http://www.mediawiki.org/w/index.php?title=Extension_talk:Scribunto/API_specification&diff=580028&oldid=559298, that I meant to tell you lot about yesterday. [00:10:22] New review: Krinkle; "Bottom line being, this enhancement should've been introduced as-is (wouldn't have passed my review)..." [mediawiki/extensions/VisualEditor] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/22701 [00:10:34] it's not even a plan [00:10:45] just an idea [00:11:38] Natterer: I think I'm not very familiar with your particular dialect of English, but "you lot" is a little dismissive or derogatory to my ears. Am I misreading? [00:12:19] Yes. It's "you (a group of people)" [00:12:40] misreading, I think [00:12:48] http://en.wiktionary.org/wiki/you_lot [00:12:58] 03(FIXED) LiquidThreads labs wiki disappeared - 10https://bugzilla.wikimedia.org/29137 +comment (10Roan Kattouw) [00:13:08] I'm asking which connotation you intend, Natterer :) [00:13:35] Out of the *1* on the page? [00:13:52] Natterer: connotation, not denotation. :) [00:13:54] connotation, not sense. [00:14:19] sumanah: AIUI it's more informal than dismissive in Commonwealth English, but we both speak American English so ... :) [00:14:19] Natterer meant to tell the people in this channel about... [00:14:28] 03(NEW) Wikidata should not return 406 Not Acceptable - 10https://bugzilla.wikimedia.org/40032 normal; MediaWiki extensions: WikidataRepo; (at.light) [00:14:43] right [00:15:13] csteipp, I was going to use $db->fetchObject instead of fetchRow but then Revision expects all the other fields to be there as well. So I'll just do $row['deleted'] = $row['rev_deleted']; before Revision::newFromRow instead [00:15:31] New patchset: Robmoen; "Plug new vector icons / styles into inspector iframe." [mediawiki/extensions/VisualEditor] (master) - https://gerrit.wikimedia.org/r/22857 [00:15:36] The dismissive is usually http://en.wiktionary.org/wiki/you_'orrible_lot [00:15:37] Natterer: You will probably want to add yourself to the cc list for this: https://bugzilla.wikimedia.org/40031 so you can track the progress of the Scribunto deployment onto English Wiktionary [00:15:46] Or http://en.wiktionary.org/wiki/you_'orrible_little_lot [00:16:03] Usually said loudly, and with a Welsh accent. [00:16:11] Natterer: ok. I read a Harry Potter fanfic where Hermione was saying "you lot" in a derogatory way but of course context is king! [00:16:21] Krenair: I think that should work. Add Aaron to the review list-- he would have the best idea of what's going on with it [00:16:32] It works, I've tried it. [00:16:35] kaldari: i cloned the repo as per your mail, made my changes as per code review, did a commit and then did git push origin but nothing happened, now git status says 'nothing to commit', how can I upload my updated code? [00:16:42] But it's just a terrible hack for something that imo shouldn't be an issue [00:16:51] I agree [00:18:36] akshayagarwal: best thing to do is install git review... [00:18:52] Anyway, I'm still trying to find where I read about the French Wikisource that wasn't your user page or the French Wikisource. I've been reading a lot over the past few days. Browser history is a wonderful thing, but it isn't half a lot of stuff to go back over. [00:19:11] i did install it while cloning the repo but now I am not sure how to push my changes [00:19:21] kaldari: ^ [00:19:39] after you do your commit, just run 'git review' and it should take care of the push for you. [00:20:45] you may have to run 'git review -s' the first time in a new repo [00:21:19] to set it up, but usually it will do this for you if you haven't already once you run git review [00:21:34] Natterer: ok, I have replied at https://www.mediawiki.org/wiki/Extension_talk:Scribunto/API_specification#Enabled.3F and quoted what Tim just said :) [00:21:53] kaldari: ok, trying now [00:22:25] if it doesn't work, we can tweak your .gitreview settings [00:22:41] New patchset: Subramanya Sastry; "Deal with missing tsr info on references token." [mediawiki/extensions/Parsoid] (master) - https://gerrit.wikimedia.org/r/22859 [00:23:02] New patchset: Krinkle; "(bug 39753) UI restores #toc as inline-block instead of table." [mediawiki/extensions/VisualEditor] (master) - https://gerrit.wikimedia.org/r/22701 [00:23:59] akshayagarwal: There's some documentation here: http://www.mediawiki.org/wiki/Git/Workflow#Push_your_change_set_to_Gerrit [00:24:20] New review: Asher; "Adding ll_title to the unique `ll_from` index will increase the cost of inserts quite a bit. This is..." [mediawiki/core] (master); V: 0 C: -1; - https://gerrit.wikimedia.org/r/17912 [00:24:56] kaldari: thats going to be quite useful, thanks! [00:25:45] also akshayagarwal_ you are probably really conversant with Git now, but I thought you might enjoy reading http://geekchick77.dreamwidth.org/2360.html [00:25:54] "Everything I wish I�d known about git... ... But didn't know to ask" [00:26:57] sumanah: looks good! I am gonna read it [00:26:58] Hmm. I think I've found a missing '^' on wikitech [00:27:22] http://wikitech.wikimedia.org/view/How_to_perform_security_fixes says git format-patch HEAD to create a patch file [00:27:35] While I had to use git format-patch HEAD^, just HEAD returns nothing [00:28:22] New patchset: Reedy; "TEST commit" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/22860 [00:28:30] Right [00:28:34] Krenair: Let me fix that [00:28:43] Change abandoned: Reedy; "(no reason)" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/22860 [00:29:03] Putting "French Wikisource" into Chrome's history brings up [[w:en:Outline of Florida]] for some reason. I'm pretty sure that there wasn't a discussion of Scribunto there. [00:29:08] ha [00:29:34] New patchset: Akshay; "1) SignupAPI.php -> updating extension creds, removing explicit setting of config variables and instead allowing them to be set up in LocalSettings 2) ApiValidateSignup.php -> updating description 3) SpecialUserSignup.php -> replaced global variables with" [mediawiki/extensions/SignupAPI] (master) - https://gerrit.wikimedia.org/r/22861 [00:29:46] New patchset: Reedy; "So, I thought we were already on 1.8.11....." [mediawiki/extensions/SyntaxHighlight_GeSHi] (master) - https://gerrit.wikimedia.org/r/22862 [00:31:13] 03(NEW) Allow javascript modules to register themselves as suitable for mobile - 10https://bugzilla.wikimedia.org/40033 normal; MediaWiki: ResourceLoader; (jrobson) [00:31:45] New review: Reedy; "I'm confused. Wasn't this done already?" [mediawiki/extensions/SyntaxHighlight_GeSHi] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/22862 [00:32:23] kaldari: pushed successfully :) have made most of the amendments as suggested by csteipp & some as suggested by you, shall make the remaining suggested changes tomorrow, thanks [00:32:37] yay [00:32:53] New review: Trevor Parscal; "Looks cool, but can we use ve_dm_branchNode instead of VeDmBranchNode so it looks more similar to th..." [mediawiki/extensions/VisualEditor] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/22700 [00:33:34] you might want to ping S Page, I think he was doing some related work [00:34:24] Hah! The 'Sourcians had the same understanding as the 'Narians: http://fr.wikisource.org/w/index.php?title=Wikisource:Scriptorium/Ao%C3%BBt_2012&diff=3609410&oldid=3609409 [00:35:24] " ... deux jours sur mediawiki.org, test.wikipedia.org et test2.wikipedia.org avant d’être installé sur tout les sites non-wikipédia." [00:36:18] kaldari: yes, I saw his test findings filed under https://www.mediawiki.org/wiki/Account_creation_user_experience/SignupAPI , have fixed a few bugs as suggested by him & have commented accordingly on https://bugzilla.wikimedia.org/show_bug.cgi?id=36225 (incidentally we both were commenting on that bug almost simultaneously without being aware of that) [00:37:01] he doesnt seem online atm so would be talking to him tomorrow [00:37:48] ori-l: Hi, have you received my mail requesting mentorship for SignupAPI [00:37:53] cool, ori-l is good to discuss with about it as well. I'm not really invovled with it (other than the code review a did a while back), so I'm sure what their current plans are. [00:38:26] yesterday I spent about half an hour trying to work out why my profiler was giving a massive timer overrun count [00:38:30] hey! akshayagarwal, yes, got your e-mail -- haven't had a chance to respond yet [00:38:50] I set a hardware watchpoint on it to detect when the overrun count changed [00:39:11] ... and noticed that you hadn't initialized it to zero? [00:39:14] and it triggered at an unrelated region of the code, where the overrun count was apparently being used as a loop counter [00:39:25] it turned out I hadn't run "make clean" after a header change [00:39:39] ori-l: no worries, shall be anticipating a positive response :) thanks [00:40:08] sometimes writing in C presents some special challenges :) [00:40:10] akshayagarwal: yup! thanks [00:40:46] Such as the challenge of naming all of the relevant headers in one's makefile. (-: [00:40:58] not my build system [00:41:00] * sumanah is back [00:41:10] New review: SPQRobin; "Agree with Nikerabbit (we've got already several functions related to language codes, and this chang..." [mediawiki/core] (master) C: -1; - https://gerrit.wikimedia.org/r/22727 [00:41:23] blame PHP and 99% of C build systems generally that don't include headers as dependencies [00:41:36] cmake solves this problem, but who would use cmake? [00:41:42] sumana, I just pointed out that the French 'Sourcerers had the same understanding as the English 'Narians. [00:41:47] http://fr.wikisource.org/w/index.php?title=Wikisource:Scriptorium/Ao%C3%BBt_2012&diff=3609410&oldid=3609409 [00:42:24] The -MD option to GCC and some text processing does, too. [00:42:28] Natterer: so, my high school French is not quite up to speed, but Google Translate tells me that Thomas PT said: "A second thing, versions 1.20-wmfXY mature, if I'm not mistaken, not two weeks but two days mediawiki.org, and test.wikipedia.org test2.wikipedia.org before being installed on all sites non-Wikipedia." [00:43:56] 03(mod) Allow javascript modules to register themselves as suitable for mobile - 10https://bugzilla.wikimedia.org/40033 +comment (10Krinkle) [00:44:10] In response to Rical saying that it would be sitting on mediawiki.org for two weeks to "ripen". [00:44:24] New patchset: Kaldari; "Beginning of page status tooltip implementation" [mediawiki/extensions/PageTriage] (master) - https://gerrit.wikimedia.org/r/22863 [00:44:26] Natterer: OK. I'm now looking at the current version of the page. [00:44:38] New patchset: Ori.livneh; "Add debug hooks to PEF2" [mediawiki/extensions/E3Experiments] (master) - https://gerrit.wikimedia.org/r/22864 [00:44:41] at least if people are upset about not getting scribunto, it indicates to me that people want it and I'm not wasting my time [00:44:51] :) true! [00:44:55] Change merged: Ori.livneh; [mediawiki/extensions/E3Experiments] (master) - https://gerrit.wikimedia.org/r/22864 [00:45:06] You've only just realized that? (-: [00:45:13] reassurance is a good thing :) [00:45:23] 03(NEW) VisualEditor: Pressing enter in a textarea should not submit the form (saveDialog-editSummary) - 10https://bugzilla.wikimedia.org/40034 normal; VisualEditor: User Interface; (krinklemail) [00:45:41] New patchset: Kaldari; "Beginning of page status tooltip implementation" [mediawiki/extensions/PageTriage] (master) - https://gerrit.wikimedia.org/r/22863 [00:45:47] TimStarling: But why not use Javascript? [00:46:00] Oy! [00:46:28] hey good idea, why didn't I think of that? [00:48:11] 03(mod) Install Scribunto on fr.wikisource - 10https://bugzilla.wikimedia.org/39744 +comment (10Sumana Harihareswara) [00:48:29] Someone will ask you why you didn't use Visual BASIC for Applications. You'll see. [00:48:58] Natterer: I've updated https://bugzilla.wikimedia.org/show_bug.cgi?id=39744 to ask the French Wikisource liaison, Tpt, to comment on their Scriptorium with the clarification that there is no deployment date/schedule for Scribunto yet [00:49:03] thank you for the reminder [00:49:14] 03(NEW) VisualEditor: Edit summary length counter confusing when in bytes - 10https://bugzilla.wikimedia.org/40035 normal; VisualEditor: User Interface; (krinklemail) [00:49:14] 03(mod) VisualEditor: Edit summary length counter counts characters instead of bytes - 10https://bugzilla.wikimedia.org/39558 (10Krinkle) [00:49:34] if you happen to know of any other communities that have been assuming/inferring/thinking that they'd get Scribunto today, I would very much like to know, Natterer [00:50:14] If I spot any more, I'll let you know. [00:50:16] you could check Uncle G's global contributions with some toolserver tool [00:50:37] I can't imagine more than one person would have found that changelog and drawn the same conclusions [00:51:22] The 'Sourcerers did. [00:51:41] Or they found something else. Because they were saying two days from August the 30th. [00:52:29] The blind leading the blind... [00:52:30] New patchset: Ori.livneh; "Update extension E3Experiments to master" [mediawiki/core] (wmf/1.20wmf10) - https://gerrit.wikimedia.org/r/22865 [00:52:33] * sumanah looks at https://toolserver.org/~luxo/contributions/contributions.php?user=Uncle+G&blocks=true [00:52:51] Change merged: Ori.livneh; [mediawiki/core] (wmf/1.20wmf10) - https://gerrit.wikimedia.org/r/22865 [00:52:55] Reedy: and this is why I'm trying to grow the tech ambassadors network and use these kinds of misunderstandings as teachable moments to get people to join it [00:53:22] It's more fun to whine and complain though! [00:53:53] * sumanah has broken her earnestness meter [00:54:30] New patchset: Subramanya Sastry; "Check if tsr is present in references tokens before generating src rt-info." [mediawiki/extensions/Parsoid] (master) - https://gerrit.wikimedia.org/r/22859 [00:54:51] New patchset: Krinkle; "(bug 39558) Implement byteLimit" [mediawiki/extensions/VisualEditor] (master) - https://gerrit.wikimedia.org/r/22866 [00:55:36] 03(FIXED) VisualEditor: Edit summary length counter counts characters instead of bytes - 10https://bugzilla.wikimedia.org/39558 +comment (10Krinkle) [00:55:36] 03(mod) VisualEditor: Edit summary length counter confusing when in bytes - 10https://bugzilla.wikimedia.org/40035 (10Krinkle) [00:56:27] New patchset: Tim Starling; "Expose Lua's time functions: os.date() etc." [mediawiki/php/luasandbox] (master) - https://gerrit.wikimedia.org/r/22867 [00:57:09] Oooooh! [00:57:50] 03(NEW) Logged out of other wikis on login to wikimania2012 wiki - 10https://bugzilla.wikimedia.org/40036 normal; Wikimedia: General/Unknown; (jasper) [00:58:01] * Jasper_Deng_sick glances at Reedy [00:58:04] yeah somehow I managed to write a few lines of code despite toddlers banging spoons all morning [00:58:49] hehe [00:58:56] Amgine, you didn't tell the MediaWikiMediaWikians about my spoon. [00:59:15] New review: Reedy; "Unfortunately this doesn't seem to be completely fixed :(" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/22685 [00:59:30] because you make too much noise with it. [00:59:41] [00:59:46] bye Amgine [01:00:20] New patchset: Tim Starling; "LuaSandbox profiler support" [mediawiki/extensions/Scribunto] (master) - https://gerrit.wikimedia.org/r/22710 [01:00:28] Change merged: Tim Starling; [mediawiki/extensions/Scribunto] (master) - https://gerrit.wikimedia.org/r/22710 [01:00:54] New patchset: Ori.livneh; "Update extension E3Experiments to master" [mediawiki/core] (wmf/1.20wmf11) - https://gerrit.wikimedia.org/r/22868 [01:01:21] Change merged: GWicke; [mediawiki/extensions/Parsoid] (master) - https://gerrit.wikimedia.org/r/22859 [01:01:23] Amgine knows about my spoon, back from when I Nattered to the Narians, the Bookies, and the Newsies years ago. [01:02:25] Hey look! The Scribuntoistificationizers have discovered orange: https://test2.wikipedia.org/wiki/Module_talk:Convert [01:02:54] New patchset: Tim Starling; "Allow time functions in LuaStandalone" [mediawiki/extensions/Scribunto] (master) - https://gerrit.wikimedia.org/r/22869 [01:03:15] Scribunto: now with orange [01:03:20] 03(NEW) File related MWTimestamp::setTimestamp : Invalid timestamp exceptions - 10https://bugzilla.wikimedia.org/40037 normal; MediaWiki: File management; (sam) [01:03:39] Change merged: Tim Starling; [mediawiki/extensions/Scribunto] (master) - https://gerrit.wikimedia.org/r/22869 [01:03:45] Hey! That's me! [01:03:48] Change merged: Ori.livneh; [mediawiki/core] (wmf/1.20wmf11) - https://gerrit.wikimedia.org/r/22868 [01:03:59] Timestamps. [01:04:15] Change merged: Tim Starling; [mediawiki/php/luasandbox] (master) - https://gerrit.wikimedia.org/r/22709 [01:04:27] ParserFunctions still has it's convert functionalioty... Not enabled (or well tested?) [01:04:43] Change merged: Tim Starling; [mediawiki/php/luasandbox] (master) - https://gerrit.wikimedia.org/r/22867 [01:05:24] Reedy: it is tested, I tested it and found that it was broken [01:05:51] Yay, bitrot [01:06:12] Can it do everything that the Pedians' one can do, Reedy? https://en.wikipedia.org/wiki/Template:Convert/doc They seem to want to be able to convert furlongs to nanometres. [01:06:37] You'd have to speak to Happy-melon (I think) [01:07:51] it can do many things, but last time I tested it, its precision calculations were wrong, it claimed that 1km = 1mi or something [01:08:38] i.e. somehow managing to get less than one significant figure of precision [01:09:18] New patchset: Ryan Lane; "Stop authentication chain if authentication fails" [mediawiki/extensions/OATHAuth] (master) - https://gerrit.wikimedia.org/r/22870 [01:09:30] Change merged: Ryan Lane; [mediawiki/extensions/OATHAuth] (master) - https://gerrit.wikimedia.org/r/22870 [01:10:49] 03(NEW) Expand Timestamp tests to test for invalid timestamps - 10https://bugzilla.wikimedia.org/40038 enhancement; MediaWiki: Unit tests; (sam) [01:11:10] wow [01:12:10] Here's that diff on frames from the Lua person, TimStarling: https://www.mediawiki.org/w/index.php?title=Talk%3ALua_scripting&diff=577853&oldid=575149 [01:12:14] New review: Reedy; "See also bug 40037" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/22685 [01:12:53] New patchset: Subramanya Sastry; "Some code cleanup / bug fixes in async signalling." [mediawiki/extensions/Parsoid] (master) - https://gerrit.wikimedia.org/r/22871 [01:13:26] he's mostly wrong [01:16:28] 03(mod) Write and implement new regex-based blacklist extension - 10https://bugzilla.wikimedia.org/16717 +comment (10MZMcBride) [01:17:06] fetching arguments takes ages, you have to expand the relevant nodes in the input dom tree [01:17:25] it was the second most expensive thing after frame:preprocess in that beatles test case [01:17:51] This heading by a Scribuntoificationist is cheering: http://test2.wikipedia.org/wiki/User:Wnt/Templates#List_of_bullshit_expressions_in_the_Lua_reference_manual [01:17:52] Is the output of your profiling public? [01:17:53] 500ns of Lua overhead for a hashtable lookup is not really relevant when you are spending milliseconds parsing wikitext [01:19:13] no, and I would have to revert all the test cases to generate it again [01:19:15] 03(NEW) Fix-up or remove ConvertParser from ParserFunctions - 10https://bugzilla.wikimedia.org/40039 normal; MediaWiki extensions: ParserFunctions; (sam) [01:20:00] but the results were not really that interesting [01:20:25] mostly the same old sources of slowness, the parser [01:20:52] I summarised my findings here: https://en.wikipedia.org/wiki/User_talk:Uncle_G#Sfn_performance [01:21:50] You told us yesterday that Lua was comparatively quite good at having large (read-only) tables with thousands of entries, as I recall. [01:22:11] yes, that's what the lua documentation says [01:22:16] Compared to having 7666 templates in the wiki, that is. [01:22:20] sumanah: https://en.wikipedia.org/wiki/User_talk:Uncle_G#Scribunto_on_en.Wiktionary [01:22:35] the lua manual says that lua can be used as a data definition language [01:22:44] Brooke: thanks, yeah, that's one of the places I have to respond to [01:22:54] maybe that is the wrong term [01:23:02] you can write data in it anyway [01:23:17] New review: Reedy; "Asher, I added this to the schema changes queue on Wikitech. I guess we should get you to do that be..." [mediawiki/core] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/17512 [01:23:35] but I'm still waiting for someone to do the testing to back that up [01:24:10] So how does one do string manipulation currently? [01:24:16] Somewhere in one of his papers, Roberto Ierusalimschy says that it's good for data-driven programs, if memory serves. [01:24:25] Just use the string module and hope there's nothing outside ASCII in your input? [01:24:39] http://www.lua.org/manual/5.1/manual.html#5.4 [01:24:43] https://en.wikipedia.org/w/index.php?title=Template:SCOTUSKey&action=edit [01:24:47] it's 8-bit clean [01:25:34] you can handle UTF-8 with single-byte operations, there are lots of places in MediaWiki where we do that [01:25:39] I'm not sure what you mean. If I pass the page title ({{FULLPAGENAME}}) to a string function, won't it screw up? [01:25:46] no [01:26:08] well, depends on what string function and exactly what you try to do to it [01:26:20] I guess string splitting will be fine. [01:26:28] And here's what that SCOTUS template does: https://en.wikipedia.org/wiki/List_of_United_States_Supreme_Court_cases,_volume_63#External_links [01:26:31] I was looking at https://en.wikipedia.org/w/index.php?title=Template:SCOTUSKey&action=edit [01:27:02] but merely passing a UTF-8 string to a Lua string function won't screw it up [01:27:13] New patchset: Subramanya Sastry; "Some code cleanup / bug fixes in async signalling." [mediawiki/extensions/Parsoid] (master) - https://gerrit.wikimedia.org/r/22871 [01:27:23] Natterer: Right. The alternative was specifying the volume number explicitly. [01:27:30] Not sure why I was so against the idea at the time. [01:27:44] Simplicity, I guess. [01:29:26] If you're thinking of Scribuntoificationizing it, you might want to look at https://test2.wikipedia.org/wiki/Module:NYRepresentatives [01:29:38] TimStarling: Unless you try to do something silly like reverse it? [01:30:09] volume = getPageName():match('volume (%d*)') [01:30:42] Didn't know a getPageName existed. [01:30:42] New patchset: Kaldari; "Page status tooltip implementation for PageTriage list view" [mediawiki/extensions/PageTriage] (master) - https://gerrit.wikimedia.org/r/22863 [01:31:41] local function getPageName() [01:31:53] Ah. [01:32:03] return current_frame:preprocess('{{PAGENAME}}') [01:32:04] end [01:32:07] local current_frame [01:32:19] function m.some_function(frame) [01:32:24] current_frame = frame [01:32:27] ... [01:32:27] end [01:33:12] Does it make any difference to avoid frame:preprocess() and just pass {{PAGENAME}} to the invocation? [01:33:16] https://test2.wikipedia.org/w/index.php?title=H%C3%BCllo&action=edit [01:33:48] https://test2.wikipedia.org/w/index.php?title=Template:NYRepresentatives&action=edit (-: [01:33:52] in terms of performance, it's about the same [01:34:17] I benchmarked frame:preprocess('{{PAGENAME}}') at 17us with a warm cache [01:34:25] and the cache is shared across all #invoke instances [01:35:33] frame.args.pagename is actually a call back to MediaWiki to expand the preprocessor dom tree where that argument is given [01:35:38] with a cache, of course [01:35:53] so it's a similar situation [01:36:06] 03(FIXED) create a new group in console of labs - 10https://bugzilla.wikimedia.org/34683 +comment (10Ryan Lane) [01:36:33] Will Parsoid speed things up when it's deployed? [01:36:39] in fact if we had a special interface like mw.page.name, then the overhead would still be maybe half [01:37:01] since you lose several microseconds just by switching from Lua to PHP [01:37:32] yes, Parsoid will speed things up, if current test cases are anything to go by [01:37:36] 14(DUP) Split sensitive configuration out of Labsconsole config, and add config to git - 10https://bugzilla.wikimedia.org/40030 +comment (10Ryan Lane) [01:37:39] 03(mod) Make the configuration file of labs console public - 10https://bugzilla.wikimedia.org/36242 +comment (10Ryan Lane) [01:37:57] That's presuming that you make mw.page a special table, rather than just populate it in C with the appropriate strings. [01:39:26] you mean have Scribunto construct a huge table of variables at the start of each parse operation, and pass it into Lua, for translation from a PHP hashtable to a Lua hashtable? [01:39:46] TimStarling: I guess my question is: will it require both Parsoid and Scribunto to be deployed to a particular wiki to get pages like "Barack Obama" to not to take so long to parse? [01:40:28] How huge is huge? As huge as this: https://test2.wikipedia.org/wiki/Module:NYRepresentatives ? Or as huge as local p = { name = {{PAGENAME}}, namespace = {{NAMESPACE}} } ? [01:41:04] we should see a big speedup without Parsoid [01:41:24] Okay. [01:41:28] 03(mod) labsconsole: error getting instance list from Nova - 10https://bugzilla.wikimedia.org/39416 +comment (10Ryan Lane) [01:41:49] Haven't you measured the articles on test2: that the Scribuntoificationists have already done? [01:42:10] 03(mod) Wikimedia Labs instances and projects pagination is funky - 10https://bugzilla.wikimedia.org/38917 +comment (10Ryan Lane) [01:42:10] Natterer: huge as in http://paste.tstarling.com/p/eztBcN.html [01:42:30] Thehelpfulone: re https://www.mediawiki.org/wiki/Wikimedia_Labs/Agreement_to_disclosure_of_personally_identifiable_information and https://www.mediawiki.org/wiki/Wikimedia_Labs/Terms_of_use and https://www.mediawiki.org/wiki/Wikimedia_Labs/Account_creation_text -- I'm following up and thanks for the heads-up [01:42:42] 03(FIXED) move the logbot to stable instance - 10https://bugzilla.wikimedia.org/36090 +comment (10Ryan Lane) [01:42:58] once you have a list, developers will reflexively add to it, there's no way to draw a line between bloat and non-bloat [01:43:02] That's far more than whan Victor has proposed. [01:43:20] everyone thinks their feature is important [01:43:25] 03(FIXED) Instances should use libnss-ldapd rather than libnss-ldap - 10https://bugzilla.wikimedia.org/34105 +comment (10Ryan Lane) [01:43:38] anyway, 17us is fast enough [01:43:42] Victor hasn't proposed including categories, or AFT5, or toolbars. [01:43:45] 5us is fast enough too [01:43:54] 03(FIXED) Fix homedirectorymanager.py to update ssh keys instead of just creating/deleting home directories - 10https://bugzilla.wikimedia.org/31949 +comment (10Ryan Lane) [01:43:58] I don't know why you want to wring your hands about whether or not it can be reduced to 1us [01:44:20] especially when the main pass will still take a few seconds [01:44:21] You're the one wringing your hands about it. [01:44:38] In fact, you're the only one who has said any such thing. [01:44:46] no, I'm saying use frame:preprocess('{{PAGENAME}}') [01:45:22] And Victor's saying use mw.page.title. [01:45:32] Or whatever it is. [01:45:43] 14(DUP) 'Configure instance' page doesn't display instance name - 10https://bugzilla.wikimedia.org/37380 +comment (10Ryan Lane) [01:45:43] 03(mod) Put the name of the instance in the Labs configure instance screen - 10https://bugzilla.wikimedia.org/35566 +comment (10Ryan Lane) [01:46:24] And you're definitely the only person who has stated that 1us straw man. [01:47:05] 1ua kind of looks like lua in this font. [01:50:30] The sensible thing to be doing is looking at all of the modules that the Scribuntoificationizers have written on test2:, and see what variables they actually have needed to use in practice. I expect that it's not much more than the local mw.page = { name = {{PAGENAME}}, namespace = {{NAMESPACE}} } that I wrote above. So all of that "huge tables" is a bit of a red herring, too. [01:51:09] Even Victor's design is overkill. [01:56:59] Nattering noise. [01:57:35] 03(mod) Enable extension Scribunto on en.WT - 10https://bugzilla.wikimedia.org/40031 +comment (10Ran Ari-Gur (User:Ruakh on WMF projects)) [01:58:03] Oooh! Let's see whether Ruakh objects. [02:00:20] 03(mod) Enable extension Scribunto on en.WT - 10https://bugzilla.wikimedia.org/40031 +comment (10MZMcBride) [02:07:57] 03(mod) Rename "chaptercommittee-l" to "affcom" (+ aliases) - 10https://bugzilla.wikimedia.org/39484 +comment (10Daniel Zahn) [02:08:01] 03(FIXED) Rename "chaptercommittee-l" to "affcom" (+ aliases) - 10https://bugzilla.wikimedia.org/39484 (10Daniel Zahn) [02:16:19] New patchset: MarkAHershberger; "Bug 39466: Fix Media: links so that hooks are called" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/20500 [02:21:52] Media wiki in on the way. [02:22:19] New patchset: Subramanya Sastry; "Added getWTSource convenience method to tokens." [mediawiki/extensions/Parsoid] (master) - https://gerrit.wikimedia.org/r/22875 [02:24:54] gerrit [02:25:12] 03(NEW) Special:PasswordReset could use some design love - 10https://bugzilla.wikimedia.org/40040 normal; MediaWiki: User interface; (b) [02:27:37] I'm heading off to dinner etc., but I'll get some more information together re Lua stuff, and thanks for bringing my attention to it, Natterer [02:27:55] thanks for the help, Amgine [02:28:01] and happy hacking, TimStarling [02:28:19] No worries. [02:29:59] revdel doesn't have an api right? [02:30:09] It should. [02:30:29] Brooke: where is it [02:31:40] liangent: Dunno. Might not exist. [02:46:39] 03(FIXED) [Regression] RTL with IE-7 (or IE 8/9 in "compatability mode") does not show any toolbar - 10https://bugzilla.wikimedia.org/34885 +comment (10kipod) [03:03:39] New patchset: Nullspoon; "Added $wgWhitelistReadRegexp for regex whitelisting" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/11137 [03:10:32] New patchset: Kaldari; "(bug 39988) Ensure time isn't just a default value before using it." [mediawiki/extensions/UploadWizard] (master) - https://gerrit.wikimedia.org/r/22879 [03:11:40] 03(FIXED) If the EXIF time for an image is 00:00:00, don't include the time in the date field - 10https://bugzilla.wikimedia.org/39988 +comment (10Ryan Kaldari) [03:16:44] 03(mod) If the EXIF time for an image is 00:00:00, don't include the time in the date field - 10https://bugzilla.wikimedia.org/39988 +comment (10Mark Holmquist) [03:17:00] 03(mod) Enable $wgCrossSiteAJAXdomains for wikimedia sites - 10https://bugzilla.wikimedia.org/20814 +comment (10Brett Zamir) [03:18:24] New patchset: Madman; "(bug 39688) action=info displays misleading edit history for redlinks" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/22880 [03:19:21] 03(mod) action=info displays misleading edit history for redlinks - 10https://bugzilla.wikimedia.org/39688 +patch-in-gerrit; +comment (10Madman) [03:19:50] New patchset: MarkAHershberger; "Bug 39466: Fix Media: links so that hooks are called" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/20500 [03:31:38] New patchset: Nullspoon; "Added $wgWhitelistReadRegexp for regex whitelisting" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/11137 [03:33:20] New review: Nullspoon; "Removed duplicate declaration of TitleTest::dataTestIsValidMoveOperation() from someone else's previ..." [mediawiki/core] (master) C: 0; - https://gerrit.wikimedia.org/r/11137 [03:37:57] New patchset: Nullspoon; "Added $wgWhitelistReadRegexp for regex whitelisting" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/11137 [03:42:47] New review: Nullspoon; "Forgot to change data source for TitleTest::testWgWhitelistReadRegexp to dataWgWhitelistReadRegexp." [mediawiki/core] (master) C: 0; - https://gerrit.wikimedia.org/r/11137 [03:55:00] 03(mod) Scribunto should support global module invocations - 10https://bugzilla.wikimedia.org/39610 +comment (10Tim Starling) [04:02:26] 03(mod) Auto-categorize Scribunto module pages that contain errors - 10https://bugzilla.wikimedia.org/39605 +comment (10Tim Starling) [04:06:20] 03(mod) Enable $wgCrossSiteAJAXdomains for wikimedia sites - 10https://bugzilla.wikimedia.org/20814 +comment (10Sergey Vladimirov) [04:08:19] 03(mod) Inconsistent state within the internal storage backends - 10https://bugzilla.wikimedia.org/39221 +comment (10Liangent) [04:13:41] 03(mod) Auto-categorize Scribunto module pages that contain errors - 10https://bugzilla.wikimedia.org/39605 +comment (10MZMcBride) [04:16:05] 14(INVALID) no support newline? - 10https://bugzilla.wikimedia.org/39556 +comment (10Tim Starling) [04:16:43] 14(INVALID) Debug console doesn't support multi-line script block - 10https://bugzilla.wikimedia.org/39566 +comment (10Tim Starling) [04:17:10] TimStarling: I find it kind of mind-boggling to think that Git <--> MediaWiki integration for Scribunto modules would be less development effort than using a wiki. [04:18:15] separation of concerns [04:18:56] Hmm. [04:19:03] it's easier to write code when you don't have to worry about other code that it might screw up [04:19:50] I'm not sure how that relates to putting Scribunto modules into a Git repo instead of using a Module namespace. (That's what you were suggesting, right?) [04:20:23] *sigh* AWS really isn't anything special [04:23:46] yes, that's what I'm suggesting [04:24:53] I don't see how a Git repo makes it easier to write code / makes it so that coders don't have to worry about other code that it might screw up. [04:25:10] Hmm, mostly the latter half. The former half is easy enough, I guess. Git was built for code. [04:27:36] New patchset: IAlex; "Do strict check for $par in Special:Log" [mediawiki/core] (master) - https://gerrit.wikimedia.org/r/22822 [04:28:59] 03(mod) Creating a Git/Gerrit/Labs account requires human intervention - 10https://bugzilla.wikimedia.org/37628 +comment (10MZMcBride) [04:29:09] so regardless of the effort taken to set it up, you don't think we should put modules in git? [04:29:22] you think it won't be good for users? [04:29:58] You lose a lot. I'm not sure what you gain. [04:30:17] the ability to change more than one module at a time [04:30:19] pre-commit testing [04:30:27] socialised testing [04:30:56] the ability to use desktop clients such as text editors, so that you aren't forced to use a web-based editor with nonfunctional scrollbars [04:31:09] Separate authentication system, no watchlist/RecentChanges integration. [04:31:28] And I'm not sure how much testing anyone is going to be doing locally. [04:31:42] Dependso on the level of complexity to set up Lua, I guess. [04:31:50] Depends * [04:31:55] it doesn't have to be locally [04:32:17] Pre-commit testing and desktop cilents suggested local testing to me. [04:32:53] If the Web-based editor has issues, those seem tangential. And we already support editing in an external editor, in theory. [04:33:37] Plus there's a huge question about whether you'd lose the advantages of using a Module namespace, isn't there? [04:34:21] Re: templatelinks for tracking, support for easy module renames, etc. [04:35:18] Plus you add a substantial overhead to participation. [04:36:22] templatelinks doesn't have interwiki support [04:36:26] That is, the barrier to entry rises considerably when it's no longer a matter of a decent Web browser but instead learning how to use and manage a Git repository (no small feat). [04:36:52] Not currently. [04:38:22] just that subproject alone, adding interwiki support to templatelinks, would be bigger than fetching modules from git [04:40:41] Maybe. But adding interwiki support to templatelinks would have a greater/more valuable impact, I think. [04:41:49] Cite doesn't appear to have tracking categories [04:42:25] I set ShortURL in apache without configuring MySQL. [04:42:36] When I visit media wiki, I see "$wgScriptPath="/w"; $wgArticlePath="/wiki/$1";" [04:42:42] It is printed on firefox. [04:42:47] Heh, I wonder if they were hacked in to the MediaWiki messages. [04:43:01] probably [04:43:18] doesn't matter, I can add them anyway [04:43:45] I don't know about parse errors, but certainly script errors from #invoke [04:44:01] "$wgScriptPath="/w"; $wgArticlePath="/wiki/$1";" is in LocalSettings.php [04:44:41] TimStarling: Parse errors seem like they'd be difficult to catch if not categorized/marked/flagged. [04:44:47] https://en.wikipedia.org/w/index.php?title=MediaWiki:Cite_croak&action=edit [04:45:01] they'll show up as script errors on the pages that invoke them [04:45:16] And more pointedly . [04:46:29] I suppose. I'm not sure I see the disadvantage to auto-categorization of parse errors, though. You're worried the category will be full of test pages or something? [04:48:27] damn [04:49:09] crocket: "Configuring MySQL"? What does MySQL have to do with short URLs? [04:49:42] Brooke, [Thu Sep 06 13:48:35 2012] [error] [client 192.168.0.79] PHP Fatal error: Call to undefined function mysql_error() in /srv/mediawiki-1.19.2/includes/db/DatabaseMysql.php on line 305 [04:50:02] You think that's related to your short URL configuration? [04:50:21] The URL to the wiki just displays the content of LocalSettings.php [04:50:29] Brooke, I don't know. [04:50:36] Short URL may not be a problem at all. [04:50:47] Displays the content of LocalSettings.php? [04:50:49] Really? [04:51:07] Is PHP installed? [04:51:20] Crocket: it sounds like php is not installed, or is not enabled. [04:51:40] Amgine, PHP is enabled. [04:51:58] I ran test.php, and it works fine. [04:51:58] crocket: How do you know? [04:52:04] What operating system are you using? [04:52:07] Ran test.php how? [04:52:18]