[00:16:20] Reedy, hey. What do you think about fixing the IRC channel names for those 4 wikis which Krinkle mentioned in the bug description? (https://bugzilla.wikimedia.org/show_bug.cgi?id=28276) [00:16:42] feel free [00:17:02] We do need to announce before we push the changes [00:17:07] OMG TEH VANDALS etc [00:17:32] I'll add a patchset to your change then [00:17:41] thanks [00:19:16] I'm not actually sure where pa.us.wikimedia.org is currently going [00:32:52] Reedy, you got rid of .org from donatewiki's one, I'd like to do that with foundationwiki as well if that sounds okay [00:33:19] I paused and thought about doing it too [00:33:22] tbh, it makes sense [00:33:58] We should either do it to both or neither... Doesn't make much sense to do only one [00:34:11] indeed [00:34:22] it's not like its useful information [00:36:33] https://gerrit.wikimedia.org/r/#/c/47307/1..2/wmf-config/InitialiseSettings.php [00:49:39] Krenair: Be very careful with those changes to irc.wm.o channel names [00:49:47] need at least a 2 month heads up [00:49:55] There's a LOT of stuff in CVN [00:50:18] and as long as wmf doesn't invest in countervandalism infrastructure, we rely on [[m:CVN]] to do it and they have a lot of decentralised infrastructure [00:50:27] (I know because I manage it partially) [00:54:11] Krinkle, makes sense. Btw, did you see https://gerrit.wikimedia.org/r/#/c/47306/ ? [00:54:30] not yet, I just logged on :) [00:54:33] Looks good to me [00:54:53] I have a similar commit in local copy that saves off about 60% of false positives currently in lqt-jshint runs [00:55:03] this eliminates 0%, but is good nonetheless [00:55:21] approved [00:55:42] https://integration.mediawiki.org/ci/job/mwext-LiquidThreads-jslint/132/consoleFull [00:55:48] A bunch more to go :) [00:55:52] Ugh, who came up with the syntax for the autopromotion settings? :( [00:55:58] Krenair: got jshint locally [00:56:01] Krenair: got jshint locally? [00:56:54] Don't think so... I've been relying on Jenkins :) [01:00:13] Krinkle, we also have https://gerrit.wikimedia.org/r/#/c/44643/ currently sitting around... [01:00:57] Krenair: Do you have nodejs installed locally? [01:01:15] Krenair: What environment are you in (mac, linux, win32, win64) [01:01:44] Reedy: just to verify - for bug 43185, you ran the maintenance script too, not just the config change? [01:01:47] linux, ubuntu 12.10 specifically [01:01:52] bawolff: Yes [01:02:04] It lied to me multiple times [01:02:11] Krenair: I'd recommend abandoning 44643 for now, too specific. I'm currently rebasing my lqt-cleanup branch, which fixes this already. [01:02:15] It's just a few line changes [01:02:21] Cool :-) [01:02:27] Which I'll let you and antoine review [01:02:31] How did it lie? [01:03:11] Krenair: 44643 is hard to review with jshint right now because of all the false positives. For all I know 44643 might introduce an error and you wouldn't notice because of the 1000s of warnings in there [01:03:36] I'd like you to review mine first so that it shaves off all of that and leaves a manageable output to work from. [01:03:49] I agree, but 44643's is hashar's... I don't like to use my LQT repo privileges much, so I'll let him abandon that himself [01:06:26] put a comment on that change [01:06:34] It doesn't have to be abandonned, it can also be rebased on top of mine in a bit [01:06:38] bawolff: It said like around 15,000 rows, but there were over 30k [01:07:21] Huh. Well that's not reassuring [01:09:30] Krinkle, what about https://gerrit.wikimedia.org/r/#/c/44635/2? [01:09:38] https://gerrit.wikimedia.org/r/#/c/44635/2 [01:10:09] Krenair: ok. So I'd recommend having nodejs in your environment. There's lots of js tools that need it. `apt-get install nodejs` I think is the ubuntu repo name for the package. [01:10:36] Krenair: any/all other commits are easier to review once the output from jshint is useful [01:11:49] I've actually just installed that, getting a weird error though... [01:13:01] Okay fixed it (https://github.com/joyent/node/issues/3911 ) [01:15:32] bawolff: I ran it 2 or 3 times till there were no more rows [01:15:43] because I had to stop it [01:15:50] but it did carry on past the original number [01:16:05] Krenair: What does `which npm` give you? (and subsequently `npm --version`) [01:17:21] `which npm` --version [01:17:32] /usr/bin/npm [01:17:42] 1.1.4 [01:18:05] Im kind of surprised its even giving a total number of rows to update on a miser mode wiki [01:22:50] $wgRC2UDPPrefix = str_replace( '\\', '', str_replace( '.org', '', $wgServer ) ); [01:27:20] Reedy: I think I know why it wasnt stopping. Its checking to see if estimateRowCount returns 0 to see if its done. But that sort of estimate usually returns 1 even if the result is 0 I believe [01:27:51] mw.org said 21333 [01:28:02] it's passed 22500 now [01:37:56] 40766 rows processed [01:44:19] Actually I misread the code. Looks like the estimating code is really inaccurate [01:53:54] About half [02:05:38] Krenair: omg, lqt is a really big mess. It's like the author's first attempt at writing javascript, ever. [02:06:04] Yup... [02:06:44] and even then, doing so with no care about leaking global variables and leaving useless statements in the middle of other code [02:06:49] like this one: [02:06:56] liquidThreads.apiRequest; [02:07:05] That doesn't do anything. [02:07:57] ok, 27 errors left. Tried to keep the diff as clean as possible. [02:08:15] a bit of self-review and diff clean up, coming your way soon :) [02:23:33] Krenair: Pushed it [02:23:34] Krenair: You can configure gerrit to highlight changes per characters instead of line, that makes that much easier to review [02:23:59] Gerrit Menus > Differences > Intraline Difference [02:24:11] Then Save it, so you'll have it next time [02:24:40] already had that on :) [02:24:45] ofcourse :) [02:28:52] https://integration.mediawiki.org/ci/job/mwext-LiquidThreads-jslint/136/console [02:28:55] 27 errors [02:28:56] B) [02:30:15] New review: Krinkle; "@Hashar: What are we waiting for? At this rate we'll have testswarm when our sun explodes." [integration/jenkins-job-builder] (master); V: 1 C: 1; - https://gerrit.wikimedia.org/r/44836 [02:31:34] brb [02:43:25] Krinkle|detached, is there any way to automatically reformat JS code? If so, is it something we could consider using on LQT? [02:43:54] Also I notice for some parts of this you change 'functionName' to functionName, removing the apostrophes... But not all places. Is this intentional? [02:52:45] * bawolff salutes your guys' bravery. I read lqt code once. There be dragons there [02:57:10] it's just PHP [02:57:22] it's automatically painful to read [02:57:31] oh apparently it's JS [02:57:36] but I retain my second statement [03:07:23] Heu. Php can be beautiful sometimes [03:07:30] *hey [03:07:55] Not always. But sometimes [03:13:14] [03:13:43] Hey amgine [03:13:56] Heya... [03:14:27] It was a bit of an ambush. I wasnt expecting anything to ding... [03:23:23] Krenair: I did it only when changing something else on the same line [03:23:25] To keep the diff clear [03:23:28] Krenair: And no, there is no stable style reformatter for javascript yet [03:23:40] jQuery is working hard on getting one for their style as well [03:23:52] when they do, we can use the same and fill in different parameters for our style guide [03:24:00] will likely be coming from the uglifyjs initiaitve [03:24:07] initiative* [03:24:21] which appears to have one of the best parsers / AST makers our there [03:24:23] out* [03:25:45] okay [03:25:57] apart from that, everything looks good [03:30:37] New patchset: Krinkle; "sphinx is now stricter" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47321 [03:30:37] New patchset: Krinkle; "Add ability to specify mavenName for maven jobs." [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47322 [03:30:38] New patchset: Krinkle; "Documentation fixes to make Sphinx happy" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47323 [03:30:38] New patchset: Krinkle; "Remove setuptools-git from setup.py." [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47324 [03:30:39] New patchset: Krinkle; "Raise exception for at least one type of syntax errors" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47325 [03:30:39] New patchset: Krinkle; "Bugfix and tidy-up" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47326 [03:30:40] New patchset: Krinkle; "Implement publisher.checkstyle" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/44836 [03:32:10] Krinkle, are you expecting me to merge this? [03:32:47] Because otherwise I'm not sure what should be done next... [03:34:54] No [03:34:59] Krenair: merge waht [03:35:11] merge what? [03:35:17] Change abandoned: Krinkle; "(no reason)" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/44836 [03:35:22] Change abandoned: Krinkle; "(no reason)" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47326 [03:35:27] Change abandoned: Krinkle; "(no reason)" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47325 [03:35:33] Change abandoned: Krinkle; "(no reason)" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47324 [03:35:37] https://gerrit.wikimedia.org/r/#/c/47316/ - JSHint: Catch some big fish. [03:35:39] Change abandoned: Krinkle; "(no reason)" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47323 [03:35:52] Change restored: Krinkle; "(no reason)" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/44836 [03:36:04] Krenair: Yes, would be nice if you could merge that one. [03:36:16] reading your comments now [03:36:26] ah, already handled [03:36:30] yeah [03:36:45] I'm sort of unsure about merging it.. I'm not that confident with JS [03:36:58] though obviously you are, and this all seems to be mass changes based on a few simple ideas [03:38:41] Krenair: the jshint link added by jenkins should be reassuring. And if you would like to know about any change in particular, I'm happy to elaborate. [03:38:45] No rush :) [03:38:56] (on irc that is) [03:39:35] ah, caught an issue [03:39:36] could be fun/useful as a teaching mechanism, though a bit awkward on my own commit. But if you'd rather not, that's okay. [03:39:54] But I'm not sure who else with merge rights will be able to do it otherwise. [03:42:27] "mw.loader.using( ['mw.action.edit']," should be mediawiki.action.edit [03:43:12] Indeed [03:43:22] case-insensitive replace, that was stupid of me. [03:43:26] mediaWiki > mw [03:43:32] mediaWiki. > mw. rather [03:43:35] but still [03:43:38] fixing.. [03:46:26] done [03:53:34] all people in the mediawiki or editor-engagement groups have merge access to LQT, although apparently it's supposed to be an EE project (https://gerrit.wikimedia.org/r/gitweb?p=mediawiki/extensions/LiquidThreads.git;a=commitdiff;h=b4a28a44d1edcbd29757889c32195843762fe5e8 ) [03:58:00] New patchset: Krinkle; "Implement publisher.checkstyle" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/44836 [03:59:39] Krenair: well, I know lots of people have merge access. But that is not by design, that's by inheritance. LQT is a complex beast and I wouldn' expect many people to tackle a commit like this that is both LQT intensive and especially also js intensive. [03:59:59] Hm.. interesting. [04:00:36] I guess that makes it easier to manage commit access. But I'm pretty sure most if not all of EE people know nothing of LQT (and aren't expected to by Terry) [04:00:43] Just for maintenance I suppose. [04:01:18] it was the only repo that didn't inherit from a non-repo access group. It needed to be put in a group, now it is. [04:01:43] New review: Krinkle; "* Rebased and fixed line-length issue" [integration/jenkins-job-builder] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/44836 [04:02:26] There's quite a few extension repos sitting around without their own groups, IIRC [04:05:05] And yes, not everyone in editor-engagement knows LQT, but the same goes for mediawiki, which gives owner access in all extension repos [04:05:42] Yep [04:06:05] But mediawiki core maintainers get access for maintenance reasons and other stuff by design. [04:06:29] Krenair: lqt doesn't inherit ee, it was explicitly added [04:07:16] yeah.. [04:07:24] I have to go now, bye [04:13:29] Susan: Why is it ambiguous for wikidata? [04:14:02] it is fixed to www.wikidata.org pretty boldly everywhere, just like any other domain name we have without a special subdomain. [04:14:33] Krinkle: The Wikidata folks want wikidata.org. [04:14:42] But currently both work without either being primary. [04:14:58] afaik www.wikidata.org is the redirect target [04:15:02] always has been [04:15:14] hm.. [04:15:18] only for http > https [04:15:28] no. [04:15:37] lol [04:15:41] No, they can't have it [04:15:41] http://wikidata.org/wiki [04:15:47] redirect loop [04:15:49] :D [04:16:00] Heh. [04:16:30] wgServer is "//www.wikidata.org" [04:16:40] There's a related bug. [04:16:51] eventhough not all non-www is redirected, mediawiki can only output links one way [04:17:08] and it outputs it as www [04:17:17] Well, I think my point was that it was going to change at some point. [04:17:30] Why ? [04:17:32] So it may be best to break everything at once. [04:17:34] All wmf domains use ww. [04:17:38] Because the Wikidata folks want non-www. [04:17:45] wikimediafoundation.org, wikipedia.org, wikimedia.org, wiktionary.org [04:17:48] etc. [04:17:57] wikimediafoundation.org is non-www. ;-) [04:18:07] https://www.wikimediafoundation.org/ [04:18:08] ugh [04:18:13] https://bugzilla.wikimedia.org/show_bug.cgi?id=41847 [04:18:31] Anyway, I think there's currently ambiguity right now about whether the IRC channel will finally be #wikidata or #www.wikidata. [04:18:44] So I'd like to see that resolved in addition to all the other breakage for consistency. [04:18:55] It doesn't make sense to break everything twice. [04:18:57] IMO. [04:19:02] Sure [04:19:09] but we can always fix it in InitialiseSettings [04:19:20] CommonSettings should only override it if InitialiseSettings doesn't override it [04:19:29] Override it? [04:19:31] (I know InitialiseSettings is loaded first, but it can be done via === false) [04:19:33] Yes [04:19:43] I'm not sure what "it" is. [04:19:51] the udp prefix [04:19:54] Okay. [04:20:13] e.g. if a project changes domains, we can fix the udp prefix to avoid changes [04:20:18] I think some wikis have multiple channels for historical reasons, too. [04:20:24] no [04:20:27] impossible [04:20:27] No? [04:20:32] never have, never can. [04:20:50] I seem to remember duplicate channels. [04:20:59] Like #commons.wikipedia and #commons.wikimedia or something. [04:21:03] Maybe I'm mis-remembering. [04:21:07] Maybe the server doesn't clean up unused channels [04:21:17] if one changes to the other, the other will be used exclusively [04:21:24] Hmm, okay. [04:21:28] we only emit one udp event [04:22:08] I'm not sure that means much. [04:22:14] maybe someone hacked something in the irc server (not in git) to redirect certain channels, but that's about it [04:22:27] Susan: The udp event string includes the channel prefix [04:22:32] #foo\t [04:22:37] So.. [04:22:40] It means everything. [04:22:44] Right. But something else could duplicate the string. [04:22:59] To send it to multiple places. [04:23:00] ... and substring replace it? [04:23:03] Sure. [04:23:11] Possible, but highly unlikely [04:23:28] The IRC feed is a (very bad) API. I think there was a lot of resistance to breaking the channels previously. [04:23:57] But, as you say, there may just be lingering empty channels. [04:24:15] Yes [04:24:18] which is why we never change them [04:24:24] #mediawiki.wikipedia [04:24:50] but the default for new ones should be improved by using the hostname instead of $lang.$site [04:24:55] $ curl -Is "http://wikidata.org/wiki" | grep Location [04:24:56] Location: http://wikidata.org/wiki [04:24:59] where $site is inferred from the db name and *wiki is wikipedia. [04:25:23] So #mediawiki.wikipedia would become #www.mediawiki [04:25:32] *after* we fix the default, yes. [04:25:35] I'm not sure if that redirect loop is a separate bug from bug 41847. [04:25:58] So first we change the default, and hardcode any special ones that aren't in there yes (basically anything that changes when we change lang.site to hostname) [04:26:05] Susan: It is a separate bug [04:26:45] Then , 1 or 2 months after an announcement has been made we can get rid of the historical overrides [04:27:47] I'm gonna file a bug about that redirect loop. [04:29:41] https://bugzilla.wikimedia.org/show_bug.cgi?id=44612 [04:30:50] Susan: https://bugzilla.wikimedia.org/show_bug.cgi?id=28276#c14 [04:33:15] Wheeeee. [07:49:51] Change abandoned: Krinkle; "(no reason)" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47322 [07:50:03] Change abandoned: Krinkle; "(no reason)" [integration/jenkins-job-builder] (master) - https://gerrit.wikimedia.org/r/47321 [19:40:24] New patchset: Hashar; "jobs for operations-debs-adminbot" [integration/jenkins-job-builder-config] (master) - https://gerrit.wikimedia.org/r/47116 [19:40:49] New review: Hashar; "rebased on latest master to get rid of a pending change. The change is already in production." [integration/jenkins-job-builder-config] (master); V: 2 C: 2; - https://gerrit.wikimedia.org/r/47116 [19:40:49] Change merged: Hashar; [integration/jenkins-job-builder-config] (master) - https://gerrit.wikimedia.org/r/47116 [19:41:53] Change abandoned: Hashar; "no tox yet, abandoning this till we get python-tox backported for Precise." [integration/jenkins-job-builder-config] (master) - https://gerrit.wikimedia.org/r/46453 [19:46:35] New review: Hashar; "Each extensions can use a different set of triggers in Zuul. For example an extension with python s..." [integration/jenkins-job-builder-config] (master); V: 0 C: -1; - https://gerrit.wikimedia.org/r/38555 [20:39:25] New patchset: Hashar; "PHPCodeSniffer jobs for MediaWiki extensions" [integration/jenkins-job-builder-config] (master) - https://gerrit.wikimedia.org/r/46769 [20:45:40] New patchset: Hashar; "Enable CodeSniffer for Translate" [integration/zuul-config] (master) - https://gerrit.wikimedia.org/r/45744 [20:45:58] New review: Hashar; "update job name to mwext-Translate-phpcs-HEAD" [integration/zuul-config] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/45744 [20:47:18] New patchset: Hashar; "Enable CodeSniffer for Translate" [integration/zuul-config] (master) - https://gerrit.wikimedia.org/r/45744 [20:47:29] Change merged: Hashar; [integration/zuul-config] (master) - https://gerrit.wikimedia.org/r/45744 [20:54:04] New review: Hashar; "Seems to be working https://integration.mediawiki.org/ci/job/mwext-Translate-phpcs-HEAD/4/violations..." [integration/zuul-config] (master) - https://gerrit.wikimedia.org/r/45744 [20:55:10] New review: Hashar; "That worked for Translate" [integration/jenkins-job-builder-config] (master); V: 2 C: 2; - https://gerrit.wikimedia.org/r/46769 [20:55:11] Change merged: Hashar; [integration/jenkins-job-builder-config] (master) - https://gerrit.wikimedia.org/r/46769