[02:26:30] !log LocalisationUpdate completed (1.20wmf11) at Mon Sep 17 02:26:30 UTC 2012 [02:26:44] Logged the message, Master [15:28:11] !log Pointing php to php-1.20wmf11 [15:28:22] Logged the message, Master [15:31:03] !log Deleted 1.20wmf10 l10n cache [15:31:13] Logged the message, Master [16:17:35] !log reedy synchronized php-1.20wmf12 'Initial sync' [16:17:44] Logged the message, Master [16:21:06] !log reedy synchronized php-1.20wmf11/extensions/ArticleFeedbackv5/ 'Bring in logging related fixes etc' [16:21:16] Logged the message, Master [16:26:04] oh, right, it's today [16:36:02] !log reedy Started syncing Wikimedia installation... : test2wiki to 1.20wmf12 and rebuild localisation cachae [16:36:11] Logged the message, Master [17:11:06] Reedy: is wmf12 live already? [17:11:15] No [17:11:25] well, maybe on test2wiki [17:11:28] waiting for scap to finish [17:13:05] * James_F has already written the VE wmf12 release notes. :-) [17:13:34] First time I've written them before Reedy's already deployed. :-) [17:15:19] Nikerabbit: are the new plural changes on wmf12= [17:15:35] s/=/?/ [the overrides] [17:22:59] I keept getting 403s checking out from git.. [17:25:13] ? [17:25:38] error: The requested URL returned error: 403 while accessing https://gerrit.wikimedia.org/r/p/mediawiki/extensions/WikimediaShopLink.git/info/refs [17:25:38] fatal: HTTP request failed [17:25:38] Unable to fetch in submodule path 'extensions/WikimediaShopLink' [17:25:38] * RoanKattouw hopes the repos haven't gotten screwed up further [17:25:45] Hmm, checking [17:25:53] I've seen it for numerous [17:26:01] just running git submodule update --init seems to make it carry on [17:26:07] is anyone still actively working on fixing repos? [17:27:02] mediawiki-config is still not consistent [17:27:06] Cloning WMShopLink WFM [17:27:12] jeremyb: How is it not consistent now? [17:27:19] I thought I'd fixed mw-config [17:29:11] Reedy: looks like we signed up for deploying wgHtml5 today....totally forgot until just now [17:29:22] !b 40107 | RoanKattouw [17:29:22] RoanKattouw: https://bugzilla.wikimedia.org/show_bug.cgi?id=40107 [17:29:43] that's still open and as of <5 mins ago i can verify it should still be open [17:29:46] Oh right [17:29:50] Those three [17:30:03] I asked Platonides to retore them but didn't hear back from him [17:30:06] Platonides: ping [17:30:07] *restore [17:30:12] right, i saw but forgot [17:30:17] maybe someone else has them though [17:30:28] Yeah, I was hoping someone would have them [17:30:35] I personally don't [17:34:09] RoanKattouw: so i'll try to hunt down Platonides. but if we end up not finding them I think we shouldn't just leave it as is [17:34:29] I think I've got them.. [17:34:30] Maybe [17:34:47] I don't have them [17:35:17] What's the easiest way to check? [17:36:04] git show ? [17:37:17] Reedy: for i in b77ecd76ea9bdd29733a7763fb6d83083d9e0ed2 51661c745ae294e736f383e4f5fdc413003b3dfc 52e7dd5c76cd4c6e673a9aa6cc7cadee9ad77fb7; do git show "$i"; done [17:37:34] nope then [17:37:34] lol [17:38:26] mm html5 yummy goodness [17:39:10] Y U NO FINISHED SCAP!? [17:39:30] * Damianz runs scap on Reedy and it finishes in 0.0000005 seconds [17:40:16] jeremyb: There's not much we can do there I suppose, other than trying to create fake placeholder commits or something [17:41:28] RoanKattouw: or just move the pointers to some other part of the tree that's not under refs/changes [17:41:43] Right [17:41:54] That would still cause weird behavior, just less weird [17:42:27] the current problem is git fetch origin refs/changes/*:refs/changes/* [17:43:48] An hour and 7 minutes!? [17:43:56] for what? [17:45:24] scap [17:45:28] and it's still not done [17:46:32] It's almost there [17:46:42] srv297-301 and snapshot1 are the ones syncing now [17:52:05] Nemo_bis: dunno [18:06:21] !log reedy Finished syncing Wikimedia installation... : test2wiki to 1.20wmf12 and rebuild localisation cachae [18:06:30] Logged the message, Master [18:06:42] An hour and a half [18:06:43] Fail. [18:06:58] it had to stop for a bathroom break [18:07:09] got distracted in the gift shop [18:07:13] these things happen [18:07:20] :-D [18:07:22] HTML5 first, then I can run scap again [18:10:37] !log reedy synchronized wmf-config/InitialiseSettings.php 'Enabled html5 on all wikis' [18:10:47] Logged the message, Master [18:13:24] !log aaron synchronized php-1.20wmf12/extensions/timeline/Timeline.php [18:13:33] Logged the message, Master [18:18:22] Good morning all, can someone ping me about a email forwarding issue? (This is Chip, btw) [18:20:46] !log reedy Started syncing Wikimedia installation... : Rebuild localisation cache [18:20:56] Logged the message, Master [18:21:13] cndiv: you mean foo@wikimedia.org -> foo@gmail.com ? [18:21:27] what exactly of html5 can make old bots break? [18:22:05] cndiv: generally (but maybe not for every single case) it's a good idea to just announce your problem to the world and then people know who to poke about it [18:22:15] yes, that's the type of change I need, just to add an additional address to an alias [18:22:28] wasn't sure how specific I can be on a public channel :-) [18:22:57] cndiv: you can be as specific as you'd like. sounds like the kind of thing that normally would go in an RT ticket. have you ever filed one? [18:23:28] yes, and normally I'd make another one but this is a little more urgent and also sensitive [18:23:55] I need to add a new address to an alias on mchenry [18:32:34] giftpflanze: just everything? [18:32:54] it depends on how broken a bot is, everything is unreliable when screenscraping [18:33:06] ok, thx [18:33:32] giftpflanze: what bots do you use/are interested in? [18:34:07] de:user:ArchivBot [18:34:17] not mine [18:34:27] cndiv: i've poked a little, will do some more in 20ish mins when I'm back. (unless it was addressed by then) [18:34:43] I think Leslie is helping me, but thanks [18:34:54] aha, silent helping [18:35:17] i got it [18:35:27] hehe didn't want people knowing i am signed in and looking ;) [18:35:42] giftpflanze: is there something that seems broken because of the HTML5 switch, or are you just being careful? [18:35:44] i noticed already anyway ;) [18:36:14] the latter [18:36:37] we'll see in about 24 minutes [18:38:44] giftpflanze: the switchover actually happened about 30 min ago [18:39:14] well, i expect the next edit then [18:47:14] !log reedy Finished syncing Wikimedia installation... : Rebuild localisation cache [18:47:23] Logged the message, Master [18:47:37] That was a lot quicker [18:47:41] 26 minutes [18:48:38] !log reedy rebuilt wikiversions.cdb and synchronized wikiversions files: testwiki and mediawikiwiki to 1.20wmf12 [18:48:47] Logged the message, Master [19:03:59] hi [19:06:06] Was mediawiki updated recently ? [19:06:14] yes [19:06:17] and no [19:06:20] frequently [19:06:36] i've spotted a weird bug I think [19:06:53] me too [19:07:12] a align=left in the wikicode is transformed in float:left in the html code [19:08:28] style="float:left;" [19:08:49] bugzilla it [19:10:44] that is intentional, I think [19:11:01] sounds like an HTML 5 thing [19:11:27] intentional yes. align=left no longer exists. [19:11:28] 17 19:10:38 < SerajewelKS> where might i find the administrator(s) of irc.wikimedia.org? [19:11:37] he should join here any sec [19:11:42] * jeremyb pasted too early [19:11:57] maybe he got lost... [19:12:11] Kyro: is it causing a problem for you ? [19:12:29] thedj[work]: we had a template that was broken because of it on wp:fr [19:12:34] Platonides: do you not have those mediawiki-config objects in your local clone? [19:12:54] Kyro: example ? because it shouldn't, unless it was already non-visibly broken :D [19:13:22] http://fr.wikipedia.org/wiki/Mod%C3%A8le:Bo%C3%AEte_d%C3%A9roulante/d%C3%A9but [19:13:32] http://fr.wikipedia.org/w/index.php?title=Mod%C3%A8le:Bo%C3%AEte_d%C3%A9roulante/d%C3%A9but&diff=83118217&oldid=79613622 [19:14:06] the align=left was transformed into style="float: left;" I messed up some pages :/ [19:14:42] it* [19:16:41] Kyro: but they are the same in term of what they mean... [19:16:55] Kyro: what was the expected behavior vs. what was it doing ? [19:17:45] For the record, the replacement is done by this part of the MW code: [19:17:45] https://gerrit.wikimedia.org/r/gitweb?p=mediawiki/core.git;a=blob;f=includes/Sanitizer.php;h=734c4ec9de319a5ac0a050178e91bdc201f20dc6;hb=HEAD#l615 [19:17:49] ah, it's an unclosed div ? [19:17:53] I did not write the template, I was just wondering why the layout was broken. I couldn't understand why align was replaced by float. [19:18:32] the bug I found: https://bugzilla.wikimedia.org/show_bug.cgi?id=40306 [19:18:53] and another one: https://bugzilla.wikimedia.org/show_bug.cgi?id=40267 [19:21:21] Kyro: it looks like there is an unmatched div. there should be at least one more {{Documentation modèle}} in the page. [19:21:58] Kyro: and if all your transclusions have this issue, then it means that your 'close' is broken as well. [19:22:04] close template [19:23:58] thedj[work]: the 3 div opened in the first template are close in the close template [19:24:03] if i'm not wrong [19:24:33] Kyro: well they are not in the template display page itself [19:24:56] but you still haven't stated what the problem is :D [19:25:17] thedj[work]: sorry I don't get what you mean [19:25:34] like visually, where is one transclusion positioned (Visually) differently than before. [19:26:43] Kyro: on http://fr.wikipedia.org/w/index.php?title=Modèle:Boîte_déroulante/début after the , there are only 2 's there should be 3. [19:27:29] thedj[work]: yeah maybe but why did we have problems on pages where theses 2 templates where included ? [19:28:11] Kyro: i don't know, you still haven't shown me where that was, so i can't make a more accurate guess. [19:29:27] http://fr.wikipedia.org/w/index.php?title=Mod%C3%A8le:Bo%C3%AEte_d%C3%A9roulante/d%C3%A9but&diff=83118217&oldid=79613622 that bit of code produced a style="float: left;" in the HTML code instead of a align="left" [19:29:47] Kyro: but that is not a bug... [19:29:52] that's intentional [19:30:17] maybe but I couldn't get why it happened [19:30:43] I first thought of a wrong css in commons.css or a crazy js … [19:30:53] because we switched to html5 and align= is not allowed anymore. [19:31:04] thedj[work]: and I didn't know that :) [19:31:12] thx for the info :) [19:33:01] https://bugzilla.wikimedia.org/40306 "May be HTML5 related?" [19:33:46] Krenair: hmm, that sucks, need to count every single tablecell... [19:35:24] Krenair: it does use align... [19:35:45] but it looks like a simple col or rowspan issue. [19:38:42] it's the aligns [19:40:10] giftpflanze: ? [19:40:51] giftpflanze: is this how it is supposed to look ? http://imagebin.org/228770 [19:41:50] yes [19:42:38] cd [19:43:09] the align on the table rows are not correctly transformed [19:44:22] what does that mean? [19:44:37] giftpflanze: that there is a bug in MediaWiki [19:44:42] ah, ok [19:45:18] you can fix locally, by using style="align:right;" on the |- elements, instead of using align= [19:45:23] but this should be fixed for sure. [19:46:53] robla: got one actual bug in our html5 transition now it seems [19:49:11] * robla is on the phone, but will ping later [19:57:02] giftpflanze: hmm, wait a minute, it's only on preview ? [20:01:56] it was reported that it is also on normal view, but for me, yes [20:03:26] thedj[work]: any details about the HTML5 issue you were mentioning ? [20:04:55] hashar: bug 40306 [20:04:59] but it's complicated. [20:05:12] why complicated [20:05:31] the attribute handler should be handling this case, but apparently it is incorrect sometimes... [20:05:54] what we first need is an easy way to reproduce it [20:06:03] the dewiki page is a bit long :-] [20:06:33] ahh comment #2 is helpful https://bugzilla.wikimedia.org/show_bug.cgi?id=40306#c2 [20:06:53] it has transformed an align into a float on the tr elements. [20:07:03] it should have transformed to text-align [20:07:23] Sanitizer.php line 644 sets that up. [20:09:22] we need some tests in tests/phpunit/includes/SanitizerTest.php ;-D [20:09:43] i think it is f ( $attribute === 'align' && !in_array( $element, $cells ) ) { on line 695 [20:10:37] apparently td align="right" gives style="text-align:right;" [20:10:40] that's it. it happily converts everything that is not in cells to a float [20:11:37] ajhhhh [20:11:40] so we need tr too [20:11:43] k got it [20:13:32] i probably should be any type of table element, with the exception of table itself [20:13:43] so caption, tbody, tfoot, tr etc. [20:14:19] so merge of $cells $tblocks and 'tr' (I think) [20:14:49] I am not sure what align on TR is supposed to do though :-] [20:14:53] just text-align? [20:15:00] yes [20:16:56] that whole block is wrong.... [20:17:19] it should ADD that, after rewriting... [20:17:37] now it overwrites. [20:17:57] thedj > https://gerrit.wikimedia.org/r/24076 [20:18:01] that is the test [20:19:36] looks good, though we probably also need some for table and div later :D [20:20:17] the problem is that this logic has secretly mixed the positioning logic into the logic that converts to content alignment. [20:22:50] hmm [20:23:03] apparently 'tr' is in the list of $presentationalAttribs [20:23:06] so that should be handled [20:23:17] hashar: no, the float hack overrides it. [20:23:27] ahh nasty one [20:23:38] see changeid I108cbd10 [20:23:44] that should be reverted and reimplemented. [20:23:50] !g I108cbd10 [20:23:50] https://gerrit.wikimedia.org/r/#q,I108cbd10,n,z [20:24:59] Next to the align -> text-align we need an align -> 'float' and then handle that case properly. [20:25:19] although that messes up the unset of course.... [20:26:50] hmm [20:27:43] the clear should only be for table i think. All other properties concern text-align. [20:28:49] that's it. the clear is only for table + align [20:29:08] eh float [20:32:44] got to go. 10:30 pM [20:32:49] :-) [20:34:29] might be best to undo the html5 change until this is fixed. stuff will all go into the page cache when people make edits. [20:35:53] na [20:35:55] will fix it up [20:35:58] just about adding tr [20:37:01] i would just apply the float only to table. [20:37:38] that will break parser tests, but I have serious doubt about those. I see no reason why

thedj[work]: the align attribute is not supported in HTML5 [20:38:43] so that will not work :-] [20:41:15] ahh [20:45:10] !log preilly synchronized wmf-config/CommonSettings.php 'enable EPUB' [20:45:21] Logged the message, Master [20:47:26] are there stats that can give with which referer people come on a certain page? [20:48:19] akoopal: I don't think they're collected in general [20:48:30] It was in a recent reader survey though [20:48:32] IIRC [20:48:41] (I inferred the former from the latter, to clarify) [20:49:14] ahh, ok [20:49:16] pity [20:49:38] a flood of ip-edits on a page, would be handy to know where they came from [20:50:02] Reedy: I got a workaround for nasty HTML5 issue [20:55:56] do we have a tracking bug for HTML5 deployment ? ;-) [20:56:03] Reedy: Ta for updating the /wmf12 in a timely fashion, I realise it's not a particularly thankful(? is that a word) task [21:03:14] ha Jarry1250 [21:03:22] we might have some issue with HTML5 since it got enabled [21:03:30] but I guess you are well aware about it already ;-) [21:03:52] I was following your chat with thedj closely [21:04:08] Is table alignment the only major thing noticed (I've got the tracking bug open in front of me) [21:04:23] great [21:04:28] will make sure we get that merged asap [21:05:27] hashar: what's the status of jenkins for self-registration? [21:05:36] Well, I'm not saying it is the only major thing noticed, but it's the only one I've yet uncovered [21:05:38] :) [21:05:48] Ryan_Lane: talked about it with rob during our 1/1 a few minutes ago [21:06:24] Ryan_Lane: will write an email to the our internal eng list about what can be done [21:06:46] Ryan_Lane: I got the gerrit/jenkins implementation finished during hack day where a +1 is needed to trigger the build. Need to setup whitelist [21:07:01] Ryan_Lane: anyway, that is what I am working on this week. Will keep you updated [21:07:11] I want self reg probably as much as you want [21:07:30] HTML5 4 life. [21:11:07] when updating that dewiki page in 49 minutes, syntax unchanged, would i introduce the weird display to all? [21:12:15] giftpflanze: whenever the page get purged, the weird stuff will show up [21:12:27] giftpflanze: we have a patch for it, will deploy it asap [21:12:43] before or after that? ;) [21:13:02] after the current conf call ;-) [21:15:54] haha :) [21:23:24] i two other people noticing the bug, you better hurry ;) [21:23:28] +have [21:37:57] giftpflanze: let them know the issue is known and we have a patch coming :) [21:39:26] i did, one was replying angrily that you (in general) weren't thinking befor doing things ;) [21:39:37] +e [21:40:13] giftpflanze: well we do peer review of all our changes [21:40:19] but sometime we get bugs ;-] [21:42:54] of course [21:44:22] they think it's far too often [22:04:28] gn8 folks