[09:10:21] New patchset: Hashar; "(bug 38882) use $wgDevelopmentWarnings" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21707 [09:11:23] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21707 [13:38:18] <^demon> hashar: I've got a jenkins question :) [13:38:31] ^demon: go ahead :) [13:39:00] <^demon> I'm trying to build 3 branches of gerrit upstream, but it only seems to be building master. [13:39:01] <^demon> https://integration.mediawiki.org/ci/job/gerrit-war/configure#section5 [13:39:16] <^demon> If I reorder them, it only seems to be building the last one. [13:39:19] <^demon> That doesn't seem right [13:40:08] oh using the git plugin [13:40:09] hmm [13:42:01] ^demon: maybe try triggering the build process several time [13:42:18] * hashar press "build now" [13:42:36] https://integration.mediawiki.org/ci/job/gerrit-war/4/ <-- stable-2.5 [13:42:47] <^demon> Ahhh, I get it now :) [13:42:58] <^demon> Silly jenkins, that's not intuitive. [13:43:14] well the build process only build the code one after each other [13:44:07] <^demon> So, I guess I should set it up to poll some $period. [13:44:17] <^demon> So it'll end up building on any new changes to the branche(s) [13:45:14] you can use the "Poll SCM" checkbox [13:46:02] <^demon> Yeah, I'll have it go once a day. [13:46:33] we have a directory holding nightly builds [13:46:38] you might want to put the result there [13:47:03] <^demon> I was going to. [13:47:04] under /srv/org/mediawiki/integration/nightly [13:47:11] <^demon> It'll public the daily war, plus a .deb (eventually) [13:47:46] jenkins is allowed to write there, so you can use a ant script / shell script / whatever to publish it there [13:48:17] it is lacking a nice web interface though :( [13:54:58] amusso@aeriale:~/projects/integration/jenkins/jobs/_shared(git:master)$ GERRIT_REFSPEC=refs/changes/20/21720/1 GERRIT_BRANCH=master GERRIT_CHANGE_NUMBER=21720 GERRIT_PATCHSET_NUMBER=1 GERRIT_PROJECT=mediawiki/extensions/InputBox ant prepare-extension -Dsourcedir=/tmp/sourcedir -Dgit.shared.dir=/tmp/gitshared [13:54:59] New patchset: Hashar; "ant: vim modeline" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21726 [13:54:59] New patchset: Hashar; "ant: gerrit-apply-change only require dir" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21727 [13:55:00] New patchset: Hashar; "Automatically import any ext in ${sourcedir}" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21728 [13:55:00] grmbl [13:55:00] New patchset: Hashar; "ant: target to apply a change to an extension and set it up" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21729 [13:55:44] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21729 [13:55:44] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21728 [13:55:44] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21727 [13:55:45] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21726 [14:00:10] pushing that on gallium [14:00:47] I need to enhance https://integration.mediawiki.org/nightly/ :-D [14:04:09] <^demon> hashar: Mind poking https://gerrit.wikimedia.org/r/#/c/21725/? [14:04:27] New patchset: Hashar; "ant: typo in PHP injection" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21734 [14:04:40] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21734 [14:05:42] ^demon: sure , will look at it [14:06:37] need to fix up jenkins first [14:07:34] <^demon> Well puppet needs to run first so the new directory exists. [14:27:07] New patchset: Hashar; "MediaWiki: cleanup extension before build" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21738 [14:29:49] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21738 [14:32:54] <^demon> hashar: I would love for us to have slave nodes one day :) [14:33:08] yup would need to set that up one day [14:33:33] but I am first working toward running tests for extensions [14:33:38] <^demon> Of course :) [14:36:05] so Apache comes with a HeaderName and ReadmeName property [14:36:16] we might have a header which is shared among all nightly directories [14:36:33] * hashar digs in http://httpd.apache.org/docs/2.2/mod/mod_autoindex.html#headername [14:37:10] grmm [14:37:17] Readme is in the footer :-D [14:44:48] hashar [14:44:50] I need u [14:44:59] <3 XD [14:45:13] I tried to set up new bastion [14:45:23] petan: deployment-bastion ? [14:45:28] but there is no link to data [14:45:29] yes [14:45:37] //usr/local/apache [14:47:01] New patchset: Hashar; "ant: dont delete the 'extensions' directory!!" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21742 [14:47:19] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21742 [14:48:25] how to make it [14:48:36] hashar ^^ [14:51:37] ^demon: did review on https://gerrit.wikimedia.org/r/#/c/21725/ [14:52:01] ^demon: you used the same pattern I have originally used for mobile nightlys [14:52:22] <^demon> Yeah, I just copy+pasted. [14:52:23] ^demon: I think it make more sense to have everything under the same directory. Aka: nightly/mediawiki nightly/gerrit nightly/mobile/wikimedia ... [14:52:38] also the README.html file is included via Apache config [14:52:54] you will have to tweak the setting so it matches /srv/org/mw/integration/nightly [14:53:00] * [14:53:05] with a wildcard [14:53:14] petan: hmm you need to apply one of the puppet class [14:53:22] how does it work [14:53:24] I want to know [14:53:26] petan: -dbdump should have it already [14:53:27] where are data [14:53:37] <^demon> hashar: I can clean it up later. Playing with gerrit now :) [14:53:53] I know I can use puppet but I want to understand it [14:54:14] which data are you looking for? [14:54:16] there is no volume mounted, so how you synchronize it [14:54:23] /usr/local/apache [14:54:27] ahhh [14:54:43] you need the nfs::apache::labs puppet class [14:54:48] I know [14:54:55] but how does it work [14:54:57] :D [14:54:59] that will set up /usr/local/apache as a symlink to some /data/project sub dir [14:55:05] eh [14:55:13] it's a symlink? [14:55:18] I must be dumb [14:55:20] have a look at manifests/nfs.pp [14:55:25] I thought I was checking that [14:55:31] it looked like a folder [14:55:43] I thought it's scap [14:55:48] we had to use a symbolic link because Gluster would not let us mount a sub directory :-( [14:55:54] hm [14:56:33] I hope you can ssh to this one [14:56:34] :D [14:59:03] hashar there are some troubles on dewiki can I run update there? [14:59:13] I think there are missing tables for AFT [14:59:16] ohhh [14:59:19] go ahead :-) [14:59:21] ok [14:59:51] petan: I can ssh to -bastion :-)))))]]]] [14:59:59] yay [15:00:00] :D [15:00:17] I submitted a puppet change to add a role::bastion class or something [15:00:23] but still pending review :( [15:00:38] wtf include_once(/apache/common/wmf-config/CommonSettings.php) [15:00:42] what is that :P [15:04:09] MediaWiki bootstrapping ? :-D [15:05:03] petrb@deployment-bastion:/home/wikipedia/common/php-master/maintenance$ php update.php [15:05:04] No MWMultiVersion instance initialized! MWScript.php wrapper not used? [15:05:09] :/ [15:06:21] ohhhhh [15:06:24] well [15:06:29] first I would use -dbdump [15:06:30] I think there is something wrong [15:06:39] ok, but is that a proper command or not [15:06:40] I am not sure -bastion works [15:06:43] where is the maintenance dir [15:06:52] I was working on that today to make it work :D [15:06:52] and when running a maintenance script you must use mediawiki multiversion [15:07:00] aka: mwscript php eval.php --wiki=enwiki [15:07:00] mm [15:07:06] aha [15:07:06] where eval.php can be any script [15:07:09] in maintenance/ [15:07:11] ok [15:07:18] so that would be: mwscript update.php --wiki=dewiki [15:07:36] mwscript.php is where [15:07:40] mwscript will find out which php version should be used [15:07:56] found it [15:08:10] it is most probably not properly setup on -bastion [15:08:20] I see it in multiversion [15:08:22] MWScript.php [15:08:45] petan: run the update on -dbdump [15:08:52] -bastion is lacking several puppet classes [15:09:00] meh [15:09:02] which? [15:09:08] I dont know [15:09:08] I really want to make it work [15:09:12] but it does lack ton of them [15:09:54] ahh [15:09:59] petan: look at https://gerrit.wikimedia.org/r/#/c/16954/ [15:10:26] specially https://gerrit.wikimedia.org/r/#/c/16954/3/manifests/role/bastion.pp,unified [15:10:48] we might need to apply another puppet class to -bastion [15:10:49] not sure which [15:11:38] fixed [15:11:40] it works now [15:12:07] how did you fix it ? :-D [15:12:18] I copied the classes from dbdump [15:12:36] yeah looks like it was missing misc::scripts [15:12:51] I need to update the role::bastion classes [15:13:23] petan: note that we need to keep -dbdump instance for now [15:13:41] don't worry I am not the deleting guy, that's Ryan [15:18:39] ;))) [15:20:57] hashar all databases need to be updated [15:21:06] there are lot of new extensions that are missing tables :D [15:21:15] no wonder everything was broken [15:21:23] who installed them? [15:21:37] I did update all the code base yesterday [15:21:50] where are logos? [15:21:56] code base [15:21:58] but not database [15:22:20] foreachwiki update.php [15:22:23] ? ;D [15:22:51] hmm [15:22:59] ok [15:23:05] you did that? [15:23:14] because now it creates some new tables on each wiki [15:23:23] so I don't think so [15:24:24] looks good [15:29:45] \O/ [15:32:33] petan: do you still need /etc from -feed ? [15:32:40] was for some IRC server iirc [15:32:48] bug is https://bugzilla.wikimedia.org/show_bug.cgi?id=38611 [15:33:28] yes [15:33:42] I replied to that bug [15:39:20] petan: maybe we should puppetize it so :) [15:39:44] ok but we can't puppetize it when we don't know how is it configured [15:39:54] I need to have it restored before :D [15:40:27] I don't remember whole config file [15:40:32] it's lot of lines [15:40:37] like 800 [15:49:12] petan: make sense. You will want to poke Ryan about it :) [15:49:24] ^demon: Gerrit does not like my evil git clone : [exec] error: RPC failed; result=22, HTTP code = 502 [15:50:01] <^demon> What'd you do? [15:51:12] ahhhh I got a bug ;-D [15:51:19] or not [15:51:20] hm [15:51:27] I am cloning mw/core from gallium [15:52:11] using the --reference parameter to reuse objects which are already presents in a local directory [15:52:16] that might be the cause of the problem [15:52:22] <^demon> Perhaps. [15:52:25] <^demon> Maybe a jgit bug [15:52:34] http://dpaste.org/GAR9u/ [15:52:50] <^demon> I'll check gerrit log [15:53:46] trying out in shell [15:53:47] git clone --reference /var/lib/jenkins/git/mediawiki/core -- https://gerrit.wikimedia.org/r/p/mediawiki/core.git /tmp/DeleteThis [15:53:54] maybe it reaches some timeout [15:54:15] we have tooooo many objects ;-D [15:54:52] <^demon> Maybe http://p.defau.lt/?4OlAmxeqZ_trXgXDUbPq4Q ? [15:55:01] <^demon> Only thing that looks plausible in gerrit's error_log [15:55:43] maybe it uploads toooo many stuff [15:56:25] <^demon> I've got zero clue here. Maybe we should file upstream. [15:56:49] remote: Counting objects: 380902, done [15:56:52] ;D [15:56:57] let me fill in a bug [15:57:10] <^demon> Ok, let me know when you've filed it. I want to CC myself. [15:58:42] waiting for the shell command to complete [15:58:52] maybe that is a timeout error while being run under Jenkins [15:59:04] <^demon> Got the same stacktrace again. [15:59:40] Receiving objects: 51% (194261/380902), 42.93 MiB | 3.10 MiB/s [15:59:40] bah [15:59:45] I dont understand --reference [16:00:02] I don't know if ti fetch the objects from remote or locally [16:00:15] it is supposed to save some disk space ;D [16:00:46] maybe I should just use a snapshot of master instead of cloning the whole repo [16:01:19] <^demon> We also probably need to clean up references on the repo. [16:01:27] <^demon> jgit is stupid and doesn't do any garbage collection. [16:01:39] <^demon> So we've probably got loose objects dangling around. [16:02:15] ^demon: https://bugzilla.wikimedia.org/show_bug.cgi?id=39737 [16:02:17] you are on CC already [16:02:26] it works from gallium under the command line [16:02:32] so might be jenkins related [16:04:09] or maybe it is just TOOOO long :-) [16:04:35] <^demon> Maybe SSH command isn't staying alive long enough :( [16:16:57] ^demon: the git command is still running on gallium :-D [16:17:29] locked in a read() call :-]] [16:26:35] out for now [16:54:58] DanielK_WMDE: thank you for reviewing blamemaps project code! [16:56:01] MichaelShavlovsk: :) thanks for working on BlameMaps! [16:56:08] MichaelShavlovsk: you might want to email wikiresearch-l about it [17:19:48] mlitn, kaldari, bsitu, chrismcmahon: Hi guys, look forward to deploying another round of revisions for E2 today -- mostly AFT5, but also PageTriage. To keep track of all this, I have started an E2 Deployment plan on Etherpad, based on Matthias's handy list of commits to AFT. Feel free to add notes and updates on that page: http://etherpad.wikimedia.org/VzkZSy0PKK [17:22:17] My understanding is that today's AFT/E2 deployment window starts at 11am PT (after SWIFT), until 3pm PT (when Mobile is expected to start deploying), according to Terry Chay. Does this work for everyone? If any member of the SWIFT or Mobile team is here now, can you confirm that works for you? [17:37:44] mlitn: Whoa, looks like CSS flipping might be seriously broken [17:38:11] Except for some reason it's only breaking for the AFT styles and not for anything else AFAICT [17:38:27] hi there RoanKattouw and TrevorParscal and rmoen and mlitn [17:38:37] hi [17:39:06] RoanKattouw: I have a bunch of tiny patches out there, just saying :) [17:39:06] TrevorParscal: it sounds like petan's OSB stuff is waiting for him, is that right? [17:39:22] Hi sumanah. [17:39:42] AaronSchulz: Add me as a reviewer and I'll review them today. Right now I'm just starting to debug a problem where RTL flipping magically turns an entire CSS file into "\n" [17:39:58] sumanah: I'm done with OSB afaik [17:40:20] RoanKattouw: you can just use https://gerrit.wikimedia.org/r/#/dashboard/16 ;) [17:40:27] heh nice thanks [17:40:44] TrevorParscal: I had a question for you at https://bugzilla.wikimedia.org/show_bug.cgi?id=32128 but it looks like Krinkle answered it [17:41:20] TrevorParscal: if you wanted to take on a little design review, there's an ext or 2 waiting at https://www.mediawiki.org/wiki/Review_queue - "Drafts" especially could use a comment from you [17:41:27] rmoen: how's your Tuesday looking? [17:41:37] i will take a peek [17:43:09] lol - Drafts… my first extension [17:44:01] RoanKattouw: sounds like you're fixin' bugs & reviewin' code today? [17:44:10] (and recovered from your weekend :-) ) [17:44:13] Yup [17:44:52] rmoen & RoanKattouw if you could do me a favor in community stuff I'd appreciate it -- notice people who have been saying useful things in code review, or submitting reasonably good patchsets, and either encouraging them to review more, or giving me their usernames [17:45:04] sumanah: Looking full for a while, Did Tchay talk with you yet about my 20% time being focused on Micro Improvements? [17:45:33] rmoen: it may be that I missed something he said to me while I was on holiday [17:45:51] rmoen: are you working on https://www.mediawiki.org/wiki/Micro_Design_Improvements ? because that sounds neat [17:46:11] I'm working with Vibha and Ironholds. Yes that is the one :) [17:47:16] rmoen: oh cool, that's such an interesting & vibrant team then, the 3 of you - wish I had a podcast of your meetings ;-) [17:49:13] sumanah, that would be interesting. [17:49:29] rmoen: are any of the https://www.mediawiki.org/wiki/Micro_Design_Improvements things that I should encourage as bitesize tasks for medium-skill developers? [17:50:13] sumanah: I would ask Ironholds about that, because some of these items are touchy [17:50:20] politically? [17:50:23] Yea [17:50:26] :/ [17:51:23] rmoen: and one more thing -- I put in a couple of review requests for you recently .... of course if you don't want to do them then you can just remove yourself as a reviewer [17:51:34] sumanah: all of the items we are addressing are being put on rationalized on village pump and Oliver is taking the heat on whatever damage we cause [17:51:56] Oliver: human heatsink :) [17:52:07] lol [17:52:26] Ironholds: First line of defense [18:03:44] mlitn, bsitu, kaldari: what's the status of the code review for page triage? Our deployment window for AFT/E2 starts now, until 3pm PT. Would it help if we marked which commits were approved and which were not on this Etherpad page? http://etherpad.wikimedia.org/VzkZSy0PKK [18:10:44] mlitn: Whoa this AFT bug is WEIRD. I ran CSSJanus::transform() on the AFT style file in eval.php on fenari and it returned nothing, but locally it works fine [18:10:58] fabriceflorin: bsitu & kaldari are currently reviewing commits - I just updated the etherpad [18:11:26] RoanKattouw: yes, makes no sense that it works fine locally [18:11:49] Yeah [18:11:55] But it's not a caching issue or anything like that [18:12:02] The CSS flipper is actually returning an empty string [18:14:40] RoanKattouw they are running the same versions right? [18:14:48] Of MW, yes [18:14:55] just saying [18:15:00] I think it might be a difference in the PHP versions or, God forbid, the PCRE versions [18:15:11] yeah, that's the next things to check [18:20:36] RoanKattouw: for what it's worth, prototype.wikimedia.org/rc-en/load.php?debug=true&lang=he&modules=jquery.articleFeedbackv5.special&only=styles also exhibits the same problem [18:20:51] ha [18:32:03] mlitn: thanks for keeping us posted on your progress. Let us know when you are ready to deploy to testing, so that chrismcmahon and I can start testing. kaldari and bsitu have just merged a few more (including the iPad version). [18:50:14] ^demon: so instead of doing git clone, I might end up doing a git archive and extract it :-D [18:50:14] ^demon: something like: git archive --remote=/var/lib/jenkins/git/mediawiki/core master | tar xf - [18:50:22] ^demon: much faster, save us from the RPC issue [19:13:26] fabriceflorin: chrismcmahon: everything's been pushed to testwiki [19:13:41] thanks mlitn [19:13:49] rebuilding i18n cache as we speak, if you see message errors, just be patient [19:14:00] other than that, all should be fully functional [19:14:23] mlitn chrismcmahon: sounds good. I'm on it. [19:30:17] looks pretty OK in IE7 so far, but really slow. also, IE7 should die in a fire. :-) [19:32:34] I do see a page error, but istr that's an artifact of testwiki, not in production [19:32:57] chrismcmahon: hehe. I have tested the protection features, and they work as intended. The only thing missing is updating the protection log to indicate what action was just taken, but I can write that as a bug for the next release. [19:33:36] Does anyone know where I could find a help page that is in the help namespace on test.wiki, so I can check that AFT now works on help pages? [19:34:20] fabriceflorin: sure search for it !! http://test.wikipedia.org/w/index.php?title=Special%3AAllPages&from=&to=&namespace=12 [19:34:35] that search for all pages in namespace Help: [19:36:16] Thanks, Hashar, much appreciated! I guess I should have known that, but am still getting up to speed on some of the technical details of the wiki universe ;o) [19:40:01] fabriceflorin: avec quelques années d'expérience tu connaitras tout les secrets de MediaWiki :-D [19:40:44] Ben oui, j'avance doucement mais surement ;o) [19:41:30] hashar: merci de toute ton aide pour établir un site beta pour article feedback ! [19:42:02] fabriceflorin: je travaille dessus en ce moment même :-D [19:43:00] C'est super! On apprécie beaucoup ton travail, qui va nous permettre de tester notre software bien mieux que maintenant. [19:48:08] fabriceflorin: chrismcmahon: you still testing, or are we good to push to production? [19:50:00] Yes, I am still testing. Lots of features, but I'm almost done with the biggest ones. Let's plan to move to production in about 10 mins. or so, if we can't find anything major. [19:50:29] Works well on iPad, though I found a bug when I try to hide my post, it leaves the Hide my post panel up on top of the gray mask hiding the post and I can't get rid of it. I can reproduce this on iPad, but not on desktop machines. Will file this as a bug, not a show stopper. [19:50:56] New patchset: Hashar; "ant: git-snapshot" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21821 [19:51:23] fabriceflorin mlitn I might have a minor IE7 issue, Activity log shows 'resolved' but does not show flag/unflag as abuse afict. fine in Chrome, about to check FF [19:51:43] New patchset: Hashar; "(bug 39742) ant: git-snapshot" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21821 [19:52:05] New review: Hashar; "PS2 adds bug 39742 in the summary" [integration/jenkins] (master); V: 1 C: 2; - https://gerrit.wikimedia.org/r/21821 [19:52:05] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21821 [19:52:27] Also, a general issue is that the "Hide my posts" actions are not shown on the activity logs, even though I would expect them to be there. It only says that I 'cleared all flags' for this post, no mention of hiding or unhiding. Not a show-stopper either. [19:52:57] fine in FF [19:53:38] so I think not a blocker, but if it still exists after deploy, I'll file a bug for it [19:53:49] fabriceflorin: is that as anonymous or as logged in user? [19:54:03] (the hide-my-post not showing up) [19:55:32] mlitn: The hide my post issue is as a logged in, confirmed editor. I am not able to use that feature as an anon, but I think that is part of the spec. Unfortunately, I don't have a non-editor registered account, or I would test that use case as well. [19:58:51] err, right, of course, can't reliably know if a feedback entry is really from a certain anon - I actually meant confirmed or unconfirmed account (where the hide my post button is placed next to "flag as abuse" rather than in the toolbox because they have no toolbox) [19:58:53] https://bugzilla.wikimedia.org/39746 <-- incidentally, "Funky Array Magic" is the new name for my band [19:59:03] you answered the question already though: confirmed editor :) [20:01:16] OK, I have completed my testing of all the key features. From my standpoint, I think we're ready to go to production. chrismcmahon, any blockers on your end? [20:02:29] fabriceflorin: other than the possible IE7 issue with flag/unflag, I'm OK [20:02:54] pushing to production - will take a while [20:03:23] chrismcmahon: Sounds good. Please add the IE7 issue to Bugzilla, if you can reproduce it. [20:03:26] marktraceur: on a scale of 1 to 10, how interested are you in browser test automation? [20:03:33] mlitn: I will create a non-editor account on production, to test hide my post for that use case. [20:03:43] fabriceflorin: I'll check it in prod first and BZ it if it persists there [20:04:10] chrismcmahon: Excellent. That makes good sense to me. [20:09:52] New patchset: Hashar; "ant: use git archive instead of git clone" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21823 [20:10:34] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/21823 [20:10:52] <^demon> hashar: You know, we could have gerrit just replicate all the repos to gallium. [20:10:59] <^demon> Then you could do a local clone and skip the network overhead. [20:11:36] ^demon: yup might need that [20:12:53] I am not sure which design I need to adopt [20:13:06] <^demon> We can play around with it at some point when gerrit's more stable. [20:27:15] chrismcmahon: Roughly 9.98 [20:27:41] marktraceur: awesome, let me drop you email about the UW test then [20:27:47] Woot woot [20:34:28] enough for today, see you tomorrow :) [20:34:50] <^demon> night hashar. [20:35:43] oh [20:35:46] before I leave [20:36:01] jeffgreen had a RPC error while doing git clone [20:36:05] he amended on https://bugzilla.wikimedia.org/show_bug.cgi?id=39737 [20:36:30] <^demon> I saw. [20:36:59] <^demon> --recursive on core is almost always going to explode. [20:37:02] <^demon> That doesn't surprise me [20:37:20] :-) [20:37:35] so I have end up writing an ant macro using git archive [20:37:50] to snapshot mediawiki @ a given revision [20:37:53] err commit [20:42:51] cya [20:52:48] fabriceflorin: chrismcmahon: it's on enwiki [20:53:39] Great timing! I just finished writing my bugs. Will test on production now. [20:54:19] fabriceflorin: on enwiki, "hide my post" seems to work for me [20:55:07] OK, I will see if I can reproduce the Hide my post issues on production. Will first start with the Protection feature, because that is the most critical right now. Stay tuned ... [20:56:28] take your time :) [21:14:07] marktraceur: sent you mail. questions welcome! [21:15:42] Woot! [21:15:46] I'll see what I can see [21:18:18] chrismcmahon: ruby-dev package appears to be required [21:18:50] fabriceflorin: when hiding my own post on iPad, closes just fine for me, can you test that one again on production? [21:19:07] marktraceur: news to me, but I believe it [21:19:33] marktraceur: checking my local... [21:19:43] chrismcmahon: Also, I get three "formal argument cannot be an instance variable" errors when running rake [21:20:29] test/browser/spec/upload_wizard/upload_wizard_spec.rb:(38|103|116) [21:21:04] marktraceur: odd, sorry about that, you're the first person to report any of that [21:21:20] * marktraceur is an edge case [21:21:28] I'll help, not to worry [21:21:59] "do | @arg |" no longer works in 1.9 I guess [21:22:06] weird [21:23:43] No, I'm on 1.8.7 [21:23:52] So I guess |@arg| isn't allowed in that [21:24:08] Replacing it with |arg| and setting @page = arg on the next line appears to work [21:24:09] marktraceur: I have 1.8.7 on OSX and it [21:24:16] and it's fine [21:24:38] * chrismcmahon multitasks [21:25:21] chrismcmahon: Can I easily run _only_ uploadwizard tests? [21:26:04] marktraceur: 'rake upload_wizard_spec' should do that [21:26:32] mlitn chrismcmahon: Good news so far on the AFT front. The Protection feature works as intended, as well as the Feedback on Help pages and the CTA6. Moving on to iPad and Hide My Post now. [21:26:37] marktraceur: I'm checking my Ubuntu env [21:26:54] fabriceflorin: I have IE7 up vs. enwiki [21:26:57] looking... [21:27:34] Cool. Keep us posted on any IE7 weirdness. [21:27:39] chrismcmahon: Apparently the login fails epicly, I'll debug and submit some patches once I get 'er working [21:28:11] marktraceur: yeah, my Ubuntu env has stopped dtrt, will investigate [21:29:14] marktraceur: actually, the bogus user/pass is causing the failure in my env. real creds would help :) [21:29:30] *nod* agreed [21:30:13] How should I set those? [21:30:36] marktraceur: to some valid value, like your real user/pass [21:30:52] chrismcmahon: In u_w_s.rb? [21:31:20] fabriceflorin mlitn I am unable to dismiss the Activity Log window in Chrome at http://en.wikipedia.org/wiki/Special:ArticleFeedbackv5 can you reproduce? [21:31:33] marktraceur: yep [21:32:25] OK, got it working with my local environment [21:32:49] "No such file or directory - /Users/christophermcmahon/Desktop/uke.jpg" <-- silly chrismcmahon [21:33:13] chrismcmahon: I can dismiss it in chrome, clicking the X in the top right corner [21:33:15] marktraceur: I TOLD you it was stinky [21:34:21] mlitn: works for me now after page reload, not sure what that was [21:36:19] fabriceflorin: Is really "Bug 39751 - Show Protect Feedback Actions in the Protection Log and My Contributions" major like in "Bug affects an area of MediaWiki that is important on WMF sites."? [21:36:30] fabriceflorin mlitn if you find (or create) a post flagged for abuse, send me a permalink? [21:36:55] marktraceur: awesome! [21:37:59] nvm, found some abuse, checking IE7... [21:38:45] chrismcmahon: http://en.wikipedia.org/wiki/Special:ArticleFeedbackv5/Golden-crowned_Sparrow/312978 [21:38:47] oh ok :p [21:40:23] mlitn: I am sorry, but I cannot seem to access AFT5 on my iPad. I keep going to the Golden-crowned sparrow page as an auto-confirmed user, but cannot see the form at the end of the page, even if I quit Safari and reopen the app. I will try now as an oversighter. [21:41:27] I assume that's cache-related, I get to see it just fine here :) [21:43:36] Weird. I can access the Golden-crowned sparrow AFT form on an iPad as a reader, but not as an auto confirmed editor or oversighter. [21:43:40] ok, flag history seems to be displaying in IE7 on prod, at least most of the time. might be a lag issue. [21:45:57] but I am getting a page error in IE7 for every enwiki page, not just AFT [21:49:05] chrismcmahon: You wrote these tests a while ago! [21:49:52] marktraceur: yes, they have not been a high priority. and I know ProofreadPage has changed, I think spans to links, since I made it. [21:51:07] :) I recognize a few bugs I fixed that made it into your tests [21:51:07] marktraceur: it says "spike" on the repo for a reason :) [21:52:03] marktraceur: my goal for that spike was more about core architecture than actual, maintainable tests. in that sense, it's a success. now to hire a QA Engineer... [21:52:52] True true [21:53:01] Making sure things work, though. Good stuff. [21:54:54] Almost there, too. One more test needs to get fixed. [21:56:11] marktraceur: I figured you'd pick this up quickly, I'm glad you're enthusiastic [21:57:29] Yay, all passed [21:57:38] Now to see if there's anything I can do to make 'er better [21:57:42] marktraceur: damn that was fast :) [21:58:11] marktraceur: I'd only worry about the UW test, the others are either deliberately bad or throwaway [21:58:15] chrismcmahon: The only real problem was that the last test didn't have the right actions....the errors fire on focus and blur, but there were no such evetns happening [21:58:27] And yeah, I'm only running the UW spec [21:58:35] ok [21:59:40] marktraceur: Selenium 2 should dtrt with fire and blur, I'd be very interested if it does not. (Se 1.0 had explicit on_blur etc calls that no longer exist) [21:59:57] *focus [22:00:47] chrismcmahon: I think because you're explicitly sending keypresses to the fields, not setting their values [22:01:18] marktraceur: cool! if you want to make a pull request for the UW spec, I'll happily merge your changes [22:01:19] For the categories field, that's the right thing, but the altitude/longitude/etc. fields don' cur. [22:02:17] mlitn: I was able to log in on iPad with a brand new account and test the new user version of 'Hide my post' (next to the 'Flag this post' link). I did not experience the same problem as when I was an editor on test (frozen flyout), but I found another bug, which is that if I try to 'unhide' my post, the flyout appears way to the left, so that I cannot use the 'Unhide' button. I will file a bug for this, but it is not a show-stopper. [22:03:29] There is a general problem with iPad, which is that the bottom line for each post wraps around on that platform, causing the particular issue reported above, as well as an overall weirdness. This may require a bit of design love, where we would shorten some of the long labels that cause the wrap-around. Will file a separate, low-priority bug for this. [22:04:18]