[00:25:28] !log Restarted Jenkins [00:25:39] Logged the message, Master [00:34:25] !log krinkle Started syncing Wikimedia installation... : Icd721011b40bb8d2c20aefa8b359a3e45413a07f [00:34:37] Logged the message, Master [00:34:57] grmpf, why isn't that one in -operations yet [00:35:39] Reedy: does domas still run them? [00:36:43] morebots? [00:36:58] I think he means logmsgbot [00:37:03] http://wikitech.wikimedia.org/view/Logmsgbot [00:38:44] either [00:39:25] TimStarling: indeed, morebots is in both already [00:40:18] wow scap is taking well over 15 minutes [00:40:26] only 15? [00:40:26] it will [00:40:46] it was about 15 minutes before that log message appeared in this channel, which is when the actual sync started [00:41:07] Took an hour and 24 minutes for me on monday [00:41:31] 1 hour 34 even [00:41:53] I see it compiles texvc on each machine, I don't know much about that but I imagine.. wouldn't that be sync'able (and compile on fenari or wherever and then copy the result?) [00:42:23] fixed. moved the bot to -operations [00:42:28] Thx [00:42:31] yw [00:42:53] not when we're running multiple distros etc [00:43:03] Aaron created a variant of scap that doesn't do that texvc stuff [00:43:19] as it's only needed on first use of new release, or when it's actually got updates [00:54:17] Reedy: right, I imagine most of the time nothing changed in texvc. The different distros is a good point though, yeah, didn't thought about that. [01:06:49] !log on brewster: updated wikimedia-task-appserver [01:07:00] Logged the message, Master [01:08:58] !log upgrading wikimedia-task-appserver on all apaches [01:09:09] Logged the message, Master [01:19:09] !log on srv233: killed frozen apt-get, running since August 17, running apt-get upgrade [01:19:17] hehe [01:19:19] Logged the message, Master [01:19:48] you see why I don't just trust puppet to deploy package updates for me [01:20:58] TimStarling: what did you change? [01:21:34] I just updated to the latest subversion head [01:21:40] I think you had the last changelog entry [01:21:51] Reedy asked me to [01:25:17] !log on srv247, srv270, srv271, srv286, mw69: killing more stuck apt-get processes, will run apt-get upgrade [01:25:28] Logged the message, Master [03:03:49] !log on es10: started mysqld [03:03:59] Logged the message, Master [07:20:27] !log upgrading luasandbox on all apaches to version 1.1-1 [07:20:38] Logged the message, Master [12:01:23] SVG creation is BROKEN! [12:01:27] http://upload.wikimedia.org/wikipedia/commons/thumb/d/da/File.svg/50px-File.svg.png [12:01:34] Error generating thumbnail

Error generating thumbnail

The source file for the specified thumbnail does not exist.

[12:02:43] anyone could fix it? [12:02:56] it seems to work some images [12:02:59] but not for all [12:03:17] it's not broken in this case [12:03:33] http://commons.wikimedia.org/w/index.php?title=File:File.svg&redirect=no [12:03:36] this is a redirect [12:03:38] yes [12:03:40] try usin the target [12:03:46] but the SVG handler should understand the redirect [12:03:51] and offer the redirect imae [12:04:20] we don't want to store thumbs of the redirected image under the original name [12:04:29] ok [12:04:30] maybe there is some other approach that makes sense [12:04:41] and gone. ah well [13:58:57] Hi [13:59:06] Any developers around? [14:09:47] Qcoder02: maybe, but if you have a question just ask it here, and wait :) [14:10:22] I was in desire of a tool like the new page Curation tool but for images [14:19:00] You should really ask in #wikimedia-dev or use bugzilla for a request like that [15:36:37] Jeff_Green: Would this apply to OTRS addresses? If so, I'm going to forward to the relevant lists. [15:36:55] !log Created wikilove tables to itwikiquote [15:37:06] Logged the message, Master [15:37:16] (RE: your last message to wikitech-l) [15:37:32] RD: which addresses do you mean exactly? [15:37:40] all of them? [15:37:47] it doesn't affect incoming mail [15:37:53] We send mail out [15:37:59] From @wikimedia addresses [15:38:12] ah--but originating at the RT server, right? [15:38:28] from https://ticket.wikimedia.org/ [15:38:43] yep, those are affected--please feel free to spread the word! [15:38:55] OK :) [15:38:56] Thanks [15:38:59] ty [15:43:37] !log Close strategywiki, zawiktionary, zh_min_nanwikibooks [15:43:47] Logged the message, Master [15:50:16] Request: POST http://www.mediawiki.org/w/index.php?title=Wikimedia_engineering_report/2012/October&action=submit, from 208.80.152.84 via sq74.wikimedia.org (squid/2.7.STABLE9) to 10.2.1.1 (10.2.1.1) [15:50:16] Error: ERR_ZERO_SIZE_OBJECT, errno [No Error] at Wed, 03 Oct 2012 15:49:53 GMT [15:50:53] (works now, but I thought I'd let you know) [16:01:41] Nemo_bis, guillom or another translation admin: Can you handle https://www.mediawiki.org/wiki/Help:Extension:Translate/transl ? [16:01:56] looking [16:02:44] Krenair: it was never marked for translation, you don't need a t.adm. [16:02:45] gerritdown [16:02:52] Nemo_bis, well apparently I do [16:02:55] serving 503 [16:03:01] And I don't really want to add flags to myself unless I know what they do :) [16:03:33] already done [16:03:38] Ryan_Lane: gerrit down [16:03:44] known [16:03:48] action=delete on that page redirected me to Special:PageTranslationDeletePage/Help:Extension:Translate/transl [16:03:53] Which said The action you have requested is limited to users in the group: Translation administrators. [16:04:01] Ryan_Lane: ETA? [16:04:32] when it's up it's up [16:04:46] it's up now [16:05:06] Nemo_bis, hm? It's still there to me [16:06:59] Krenair: job queue [16:08:25] hard deleted [16:09:03] ok, ty [19:39:17] hi [19:39:45] hello [19:39:49] what is expected deployment time for this bugfix: [19:39:51] "Chabik Anna" [19:40:00] https://bugzilla.wikimedia.org/show_bug.cgi?id=40444 [19:40:05] oops :) [19:40:19] copy and paste fail? :) [19:40:31] uhm [19:41:09] we have a lot of empty pages in main for over two weeks because of this bug [19:41:25] and no temporary workaround... [19:41:53] ankry: it's here https://www.mediawiki.org/wiki/MediaWiki_1.21/wmf1#Deployment_of_MediaWiki_1.21wmf1_to_Wikimedia_sites [19:42:21] so, later today [19:44:08] Nemo_bis: OK, thanks. [20:12:17] ankry: try now [20:12:25] well, not yet [21:58:28] hey RoanKattouw: if I want to switch the deployment train to use a new deployment branch of CentralNotice, is it as simple as changing the .gitreview on the 1.21wmf1 branch in core? [21:58:57] Ahm, no? How do you figure that? [21:59:29] Are you trying to deploy CentralNotice from some branch that isn't master? Is this branch currently behind on master? [21:59:47] we branched master. The two should be identical at the moment [21:59:50] OK [21:59:55] So what I would strongly recommend [22:00:00] (I got that from some twisted reading of "How to deploy code") [22:00:04] is to use master as your deployment branch [22:00:25] We can use Gerrit ACLs to help you lock it down as tightly as you need it to be [22:00:47] we are not concerned as much about access [22:01:01] we are concerned that changes could catch the deploy train during the fundraiser and we don't want that [22:01:06] Right [22:01:34] So ideally you just wouldn't merge things into master. But instead, we can ask Reedy to run the train using a frozen version of CN [22:02:00] Actually, I can probably modify this config myself [22:02:13] but if we did have an emergency change to make, that becomes a lot harder if master is the deployment branch [22:03:51] RoanKattouw: if you have strong feelings about this, I will let you discuss them with Katie and Kaldari. I was simply doing as told :-) [22:05:45] * RoanKattouw spots typo in the code handling pinned extensions [22:05:57] It looks like you might be able to pin an ext to a non-master branch [22:06:11] Or at least to a specific commit hash [22:06:25] pgehres: How are emergency changes hard if master is the deployment branch? [22:06:56] RoanKattouw: if there are changes that have been merged into master since we pinned it on the deployment train [22:08:26] Ah, right [22:08:36] anyway, i forwarded the email chain to you. its up to Katie and Kaldari what we do [22:08:39] equest: POST http://meta.wikimedia.org/w/index.php?title=Special:UserLogin&action=submitlogin&type=login&returnto=Special:Watchlist, from 91.198.174.55 via sq76.wikimedia.org (squid/2.7.STABLE9) to 10.2.1.1 (10.2.1.1) [22:08:39] Error: ERR_READ_TIMEOUT, errno [No Error] at Wed, 03 Oct 2012 22:08:02 GMT [22:08:39] Changes you want to push out in the next deploy [22:08:45] Now I see what you were talking about re the How to deploy code page [22:08:57] it is not the most straightforward page [22:09:40] Yeah :S [22:09:44] Anyway [22:10:16] OK, I see your argument [22:10:25] It looks like we'll theoretically be able to pin CentralNotice to something else instead [22:10:56] Or at least pin it to "whatever the previously deployed version was", which means it will never update unless you work through 3 kilobytes of [[How to deploy code]] instructions to manually update it [22:10:57] katie already does something similar for DonationInterface on the payments cluster [22:11:21] she cherry-picks into a deploy branch to deploy [22:12:32] Right [22:12:38] Yeah we support that too [22:13:13] We also theoretically support pinning extensions so they bypass the train and populate into wmfN from wmf{N-1} rather than master [22:13:24] But I just read that code and it doesn't look like it could possibly work [22:14:13] Meh no ^demon [22:14:26] i can catch him later [22:19:12] pgehres: Responded via e-mail [22:19:45] Thanks [22:57:55] I have a question about out HTML5 support [22:58:00] out/our [22:59:02] I saw someone on enwiki making a suggestion that a user should remove the font tag from their signature. Now of course font is not in HTML5, but since our archies are full of it, I assume that mediawiki will contonue to translate it out of wikitext into something that is valid html5, right? [23:00:51] carl-cbm: If the user uses foo that will just stay as is, we don't manipulate html by users (except for security reasons) [23:01:15] hoo: so if a talk page has lots of tags, we just serve it as invalid html5? [23:01:28] carl-cbm: Well, yes [23:01:30] we traditionally used tidy to fix bad html by users [23:01:51] That's not really bad html in that sense, it's bad style, but not broken [23:02:23] it's not well-formed html5, if I understand correctly they completely removed the tag rather than just deprecating it [23:03:12] carl-cbm: But browsers can display it totally fine, it doesn't really break anything except a convention [23:04:18] Interesting. I thought that the plan was for us to have well-formed html5, so that must be what I misunderstood [23:04:26] if browers are removing it someday (which wont happen anytime /soon/) we will have to handle that, but that's not anywhere near a realistic issue [23:04:46] carl-cbm: The software itself outputs fine HTML, if the users add garbage, well :/ [23:05:24] That's even worse for the MediaWiki: messages including html/ xml, those aren't even validated by tiny [23:05:25] that was not the case historically; it used to be that the only way to get the system to emit bad html was to be an admin and edit a system message that did not get passed through tidy [23:05:27] * tidy [23:05:46] right, we agree about the messages [23:06:20] but it used to be that e.g. article content was served as valid HTML even if the users put bad HTML in it [23:06:21] It depends on the definition of bad html... sure it's bad, but it's not broken [23:06:37] carl-cbm: Broken html get's fixed (eg. open tags get closed) [23:06:43] if it won't get through a validator, that's what I'm talking about. [23:07:22] anyway, if the plan is just to ignore it, that's fine. [23:08:17] carl-cbm: http://validator.w3.org/check?uri=http%3A%2F%2Fen.wikipedia.org%2Fwiki%2FBarack_Obama&charset=%28detect+automatically%29&doctype=Inline&group=0 [23:08:51] We can't fix them all automatically without probably breaking the output and wasting a lot of cpu time [23:09:03] right - the place I went wrong was assuming we would have a tidy that converts these things to HTML5, but apparently that is where I went wrong. [23:09:52] No, we only have a tidy which fixes the things that would break stuff (eg. unclosed tags), but we don't have the resources nor the libraries to clean up according to all conventions [23:10:48] hoo: some of them seem weird though, like the "language subtag" ones. for example "vls" it complains about as not a valid ISO language part.. but then when you follow the link to IANA right below, "vls" exists in the list [23:11:40] mutante: Sure, the validator isn't perfect, but it's a good point to start [23:11:57] most of the errors are probably perfectly valid [23:12:06] yeah