[00:14:18] hey does anyone have experience using the External Data extension to media wiki? [00:33:42] You want #mediawiki really for that, and/or the semantic channels [00:33:49] Yaron is in #mediawiki [00:34:49] Will someone please review https://gerrit.wikimedia.org/r/#change,6049 ? Fixes a somewhat common fatal we're seeing... I'll merge it later [03:33:30] $ ssh srv270 'echo $TERM' [03:33:30] dumb [03:33:31] $ ssh srv270 'sudo -u mwdeploy sh -c '\''echo $TERM'\''' [03:33:31] unknown [03:33:32] curious, yes? [03:58:28] /* Provide default values for $TERM and $PATH if they are not set. */ [03:58:29] if (!ISSET(didvar, DID_TERM)) [03:58:29] sudo_putenv("TERM=unknown", false, false); [07:23:45] I see that /home/wikipedia/common/php-1.20wmf2/LocalSettings.php in a reversion from 1.20wmf1 now expects all mw install hosts to have /home mounted again [07:23:52] LocalSettings.php: include_once( "/home/wikipedia/common/wmf-config/CommonSettings.php" ); [07:23:56] no if, no nothing [07:24:39] I have no idea where this file is maintained any more, the directory is git based but there's no git log for that. so... [07:25:44] I see reedy owns it, maybe I'll try nagging him [08:43:28] nebermind, I actually pushed out a changed copy (identical in fact to 1.20wmf1/LocalSettings.php :-P) [08:43:42] seems like the job runners might have gotten bit with this. I am stil checking [13:51:49] Hi. For a file such as Full resolution ‎(3,456 × 2,304 pixels) http://commons.wikimedia.org/wiki/File:CTBTO_International_Scientific_Studies_2009_-_day_1_-_Flickr_-_The_Official_CTBTO_Photostream_(26).jpg [13:52:09] is it possible for me to get the value 3,456 and 2,304? [13:52:18] through something like magicwords [14:30:49] I don't really have troubles with anything funny, but shouldn't we replace http://noc.wikimedia.org/conf/highlight.php with list of files which you can actually see? :P [14:31:00] I did the same on beta [14:31:16] so when you make a type you could check if name of file is correct [14:31:20] typo [14:31:40] I meant did a typo - s/type/typo [14:31:49] I shouldn't write today [14:32:40] :| [15:07:09] ^demon: regarding PHP 5.3 in https://gerrit.wikimedia.org/r/5021 [15:07:32] I am going to reply to tim that 5.3 offers closure and namespaces :D [15:07:55] <^demon> Namespaces would be cool if PHP hadn't adopted such batshit syntax for them ;-) [15:19:46] ^demon: syntax is probably not perfect, yet will be great to have the functionality [15:19:57] <^demon> Perhaps. I'm not sold. [15:19:58] will have to convince Tim though :-] [15:20:34] <^demon> I'm in Tim's boat I think. 5.3 is ok but nobody's managed to convince me that the "awesome features" it adds are really all that awesome. [15:25:48] ^demon: same there, still think closure can helps in some corner cases [15:26:53] same here [15:29:10] <^demon> Closures make writing one-line callbacks easier...that's about it. I don't know the performance implications though. [15:44:49] Hi. For a file such as to extract 3,456 and 2,304 value from a 3,456 by 2,304 pixel file? Perhaps through something like magicwords. [15:45:18] *Hi. I'd like to extract the 3,456 and 2,304 value from a 3,456 by 2,304 pixel file? Perhaps through something like magicwords. [15:46:10] * sumanah wonders whether it matters what the file format is? bmp? [15:47:10] its the value wiki already shows [15:47:18] http://commons.wikimedia.org/wiki/File:CTBTO_International_Scientific_Studies_2009_-_day_1_-_Flickr_-_The_Official_CTBTO_Photostream_(26).jpg [15:47:24] the numbers appear below the image [15:47:25] ToAruShiroiNeko: you will have better luck in #mediawiki I think [15:47:45] <^demon> What's a "3,456 and 2,304 value?" [15:47:51] its the file resolution [15:48:07] Full resolution ‎(3,456 × 2,304 pixels, file size: 3.45 MB, MIME type: image/jpeg) [15:48:31] <^demon> Ahh, so you want a parser function to get that resolution? [15:48:35] yes [15:48:44] I want to compute if it is a widescreen image or not [15:48:50] or if it is large enough [15:48:56] <^demon> There isn't one afaik. [15:49:01] right [15:49:24] there also is no simple way to get the file extension as far as I know. [15:50:45] <^demon> There's probably some atrocious string function that will get you the file extension. [15:51:10] Wikimedia sites do not have string functions enabled so I am unsure if I can do that [15:51:32] it would make my life simpler if I could just get it through a magicword like {{SUBPAGENAME}} [15:51:46] {{FILEEXTENSION}} maybe [15:52:31] magicwords could be: {{FILEWIDTH}}, {{FILEHEIGHT}}, {{FILESIZE}}, {{FILEEXTENSION}} [15:53:55] file namespace is a bit special :) [15:54:39] <^demon> Wikimedia sites wrote their own awful string functions. [15:54:56] <^demon> {{padleft}} for example. [15:56:35] I want to avoid those to be honest [15:56:41] they slow things down significantly [16:01:30] I want to for example exclude pdf, ogg, ogv, oga files from consideration of wallpapers [16:02:11] or I want to only include png, jpg, gif only if their resolution is above a certain threshold [16:31:17] is there a way to get the filesystem path where an uploaded file resides? [16:32:18] i'm looking around mediawikis' file/repository architecture code and no obvious method is popping out to me [16:55:56] http://upload.wikimedia.org/wikipedia/commons/thumb/1/1d/Crater_-_Flickr_-_The_Official_CTBTO_Photostream.jpg/640px-Crater_-_Flickr_-_The_Official_CTBTO_Photostream.jpg [16:56:08] I am having issues with the changes to files [16:56:57] I believe issue may be related to https://bugzilla.wikimedia.org/show_bug.cgi?id=34041 [17:15:57] Hi 20% people. gwicke how was your day? [17:16:21] sumanah: I worked on the bug you asked me to look at yesterday [17:16:22] hi Aaron|away - how are you doing? [17:16:27] gwicke: great, thank you [17:16:28] it has nothing to do with parsing though [17:16:45] yeah, it was robla who thought you would be able to work on it [17:17:12] remind me of the bug #? [17:17:34] https://bugzilla.wikimedia.org/show_bug.cgi?id=31576 [17:18:01] it seems to be caused by 33409 [17:18:05] * Aaron|away does ok [17:18:10] hi preilly - am I right in predicting that you'd like to take today to work on the Git tutorial? [17:18:31] which will be easier to patch with PHP 5.4, but that would not be short-term either [17:18:34] * RoanKattouw also wonders why that was assigned to Gabriel [17:18:42] cause Tim seemed to understand why it was happening [17:19:13] hexmode: I might not quite be understanding, can you tell me about the priority level of https://bugzilla.wikimedia.org/show_bug.cgi?id=36339 ? [17:19:37] looking at the bug, sumanah [17:19:43] thanks hexmode [17:20:21] since Tim worked on it, error reporting was marginally improved in PHP [17:20:44] AaronSchulz: do you have plans for today? [17:21:03] robla: since this week's deployments, maybe you know of something urgent for Aaron to work on [17:21:04] <^demon|away> gwicke: 5.4 they introduced better syntax error reporting right? So you don't end up with "Unexpected T_SOMEWEIRDTOKENYOUHAVETOLOOKUP" [17:21:32] * AaronSchulz wonders why gerrit is about as fast as parsing a long article [17:21:40] Tim-away: ping [17:21:48] ^demon|away: the cdb close code path misses error checking in many places [17:22:03] sumanah: I see a few revs to CR [17:22:13] <^demon|away> AaronSchulz: manganese is in Virginia, db9 is in florida. [17:22:15] AaronSchulz: ok, that's great too [17:22:25] my impression is that it might now be possible to fix it by just hacking PHP 5.4 DBA bindings [17:22:26] sumanah: "After a global right was removed i got a mail that my user page was created (although this already exists and is unchanged)" is clearly a bug and very confusing. If you think "HIGH" priority isn't appropriate, I understand, but I think sending out completely unrelated emails is something that should be fixed quickly. [17:22:30] ^demon|away: I know, but still, is it not doing batch queries? [17:22:50] hexmode: aha! ok, I found the summary a little confusing, will reword [17:23:04] <^demon|away> AaronSchulz: Well it might be a tiny bit slow because I flushed all the gerrit caches like ~15m ago ;-) [17:23:20] <^demon|away> So it's a bit cold. [17:23:54] hexmode: am I right in inferring that this should be a pretty easy fix? [17:24:00] ^demon|away: it is not about syntax errors though [17:24:09] preilly: ping [17:24:24] sumanah: I think you're right but I'm not intimately familiar with that bit of code. [17:24:35] <^demon|away> gwicke: I wasn't aware on the background of your discussion. I was just commenting on "better error reporting" :) [17:24:46] sumanah: hence my suggestion for 20% ;) [17:25:12] ^demon|away: k [17:25:12] yep [17:25:42] gwicke: I had forgotten the full history of 31576...sorry. That one might end up being a good 20% project for preilly ;-) [17:27:16] robla: if feasible, increasing the size of /tmp partitions would likely avoid this problem completely [17:27:30] hexmode: re https://bugzilla.wikimedia.org/show_bug.cgi?id=36339 (wrong mail template used: after global right changes, mail incorrectly reports page creation) .... about how many people would be affected, every day? [17:27:35] running out of disk space might not be so nice for other things either [17:27:54] yeah, I was thinking about that too. we've been pressing on the repartitioning project already [17:27:59] * robla looks up RT ticket [17:29:18] sumanah: I don't know that, but I can find out... 1m [17:29:28] Thanks [17:30:14] <^demon|away> RoanKattouw: Can I rename AFTv5-omniit to extension-ArticleFeedbackv5-omniti for consistency? [17:30:50] rt ticket regarding repartitioning: http://rt.wikimedia.org/Ticket/Display.html?id=2524 [17:31:15] robla: I believe I don't have an account there [17:31:43] * robla hates RT [17:34:23] * robla goes offline for a bit [17:34:55] gwicke: see https://bugzilla.wikimedia.org/show_bug.cgi?id=36488 [17:36:04] I suppose that doesn't help us on l10n cache build, though, since that runs on fenari(?) [17:36:14] anyhow...actually going now [17:37:16] It is now [17:37:20] It wasn't on fenari before [17:37:41] ok, thanks all [17:37:42] <^demon> <^demon>: RoanKattouw: Can I rename AFTv5-omniit to extension-ArticleFeedbackv5-omniti for consistency? [17:38:51] ^demon: Yes [17:39:03] <^demon> Mmk thanks. [17:39:44] <^demon> We really need to finish skinning our gerrit once we update to 2.3. I'm tired of forgetting where I'm on our gerrit or gerrit's gerrit. [18:47:34] robla: ping [18:47:47] hi gwicke...what's up? [18:48:01] I am wondering if there is a good reason for those partitions anyway [18:48:19] filesystems allow to reserve some space for root [18:48:28] and log rotation can be based on space used [18:49:05] so why not use a single partition? [18:49:41] the RT ticket has a lot of the dialog. Mark B tells us that the existing partitioning is that way purely for legacy reasons [18:50:21] k [19:48:28] Hi...does any one tried wikieditor on mediawiki 1.19.0rc [19:48:29] ? [19:48:53] did* [20:59:57] RoanKattouw_away: you around? [21:00:23] AaronSchulz: howdy [21:00:45] AaronSchulz: does mediawiki automatically serialize things, or do we need to do so ourselves? [21:00:54] for the database [21:01:55] what things? [21:02:14] <^demon> Like on $db->insert()? No, it doesn't serialize anything. [21:02:21] ok [21:02:36] so I need to run serialize on the data, then unserialize when I fetch it [21:02:38] yeah, the DB layer does not automatically serialize anything [21:02:49] * Ryan_Lane nods [21:02:52] good to know. thanks [21:02:55] <^demon> Stuff sent to $wgMemc is automatically serialized though :) [21:03:19] yeah. that's why I asked :) [21:03:35] I haven't needed to deal with serialized data in the database till now, so it's a good time to ask. heh [21:04:45] <^demon> Oh, I fixed 5810 for you :) [21:04:59] 5810? [21:05:09] <^demon> !g 5810 [21:05:09] https://gerrit.wikimedia.org/r/#q,5810,n,z [21:05:09] https://gerrit.wikimedia.org/r/#q,5810,n,z [21:05:11] <^demon> gitweb stuff [21:05:20] ah [21:05:20] ok [21:06:09] ^demon: merged [21:06:21] <^demon> Thanks [21:08:28] woha [21:08:29] http://pastebin.com/Q3DqVFh8 [21:08:36] mediawiki blew in my face!? [21:08:58] LBFactory_Multi::newExternalLB: Unknown cluster "cluster23" [21:14:25] ToAruShiroiNeko: see #wikimedia-tech [22:42:27] crap, when was CryptRand.php added? [22:42:42] 1.19? [22:44:31] bah. 1.20 [22:46:17] AaronSchulz: how do I get a checkout of core and all the extensions used in production? [22:48:38] Ryan_Lane: I assume you have the core clone [22:48:42] I do [22:48:48] I need to add all of the submodules? [22:49:04] Ryan_Lane: Just switch to the wmf/1.20wmf2 branch [22:49:34] unless you want the master repos [22:49:51] <^demon> Ryan_Lane: Did you push out 5810 after you merged it? [22:49:56] ^demon: yes [22:50:01] I didn't force a puppet run [22:50:11] <^demon> It said puppet ran 8m ago. [22:50:15] <^demon> But I can't find the file I added :\ [22:50:22] ah. sweet [22:50:26] the branch has the extensions [22:50:49] so, if I want to put some more extensions underneath this... [22:50:52] just clone them? [22:51:02] <^demon> You can add them as submodules. [22:51:15] ah. cool [22:51:21] <^demon> `git submodule add ` `git submodule update --init` [22:51:55] <^demon> submodules are coooool :) [22:52:35] Wouldn't that leave you with permanent local changes though [22:52:39] unless you gitignore it or somethig [22:53:46] meh [22:53:49] I can ignore it [22:54:31] hm [22:54:45] doesn't seem to be helping [22:58:06] I'll just keep them as untracked [23:02:14] Magic words such as {{FILEWIDTH}}, {{FILEHEIGHT}}, {{FILESIZE}}, {{FILEEXTENSION}} would be helpful to me. [23:02:33] How difficult is it to create magicwords? [23:03:00] https://www.mediawiki.org/wiki/Wikimedia_engineering_20%25_policy#IRC_checkins tells me that I should now check in with kaldari, Tim-away, bsitu, and awjr [23:03:27] * awjr waves [23:03:58] hi awjr! is tomorrow another dippy-bird day? [23:04:04] preilly: ping re Git tutorial [23:04:31] sumanah: no, i think dippy bird is where i want it now. i think i'd like to work on expanding the HTMLForm functionality [23:04:38] to make it less restrictive. [23:04:51] awjr: pointer? I am curious [23:05:17] the HTMLForm class in MW makes it pretty straightforward to programatically create MW-friendly forms [23:05:33] but, it kinda ties your hands in terms of how the forms get rendered [23:05:50] it makes it so that forms and form elements get displayed inside of table elements [23:05:54] ^demon: so, the directories for the submodules are here [23:05:54] which kinda sucks. [23:05:59] ^demon: how do I init them? [23:06:15] i'd like to make it more flexible so you can have HTMLForm otuput forms wrapped in tables, divs, or just return raw form elements [23:06:19] <^demon> How did the directories get there? [23:06:39] it's been on my todo list for a while and just haven't had the chance to get to it [23:07:14] awjr: I am at the end of a longish day and thus am probably not very good at inferring how this Benefits The Larger Community etc [23:07:21] god damn it [23:07:29] the submodules use ssh!? [23:07:37] -_- [23:07:48] * Ryan_Lane is going to get stabby [23:08:58] sumanah: well, it makes a piece of MW that kinda sucks suck a lot less and be more usable. right now people who wish to make forms from within their extensions can chose between building them with HTMLForm (very MW friendly, but very restrictive) or building themselves (more error prone, more prone to security vuln, but very flexible) [23:09:17] so imho it would greatly benefit the larger MW developer community [23:09:26] <^demon> Ryan_Lane: They use whatever url you gave them. [23:09:38] no [23:09:47] it's submodules that exist in the branch [23:09:48] RoanKattouw_away: Our submission for Wikimania got accepted [23:09:50] (I didn't get any email about it, noticed it on the wiki) [23:09:51] in general i believe that making MW more awesome generally benefits the larger community... [23:09:59] my top level branch is httpos [23:10:13] <^demon> Ryan_Lane: Since we're only pulling in changes and not pushing things back to the submodules' repos, we could switch that to https. [23:10:25] that would make sense. [23:10:27] awjr: cool! no prob, sorry [23:10:32] in production we expect people to use ssh? [23:10:34] that's odd [23:10:35] awjr: I just do not know enough about that form stuff [23:10:36] no worries :) [23:10:43] Krinkle: Yeah, sorry, forgot to tell you [23:10:44] <^demon> I'll adjust make-wmf-branch so 1.20wmf3 will use https. [23:10:55] awjr: you know who would be interested in knowing of your work? [23:10:57] Edit was made April 30 [23:11:04] I was notified like last week [23:11:05] awjr: the GSoC student who is working on a Convention extension [23:11:35] sumanah: i think i worked with him in pune... what's his name again? [23:11:42] https://wikimania2012.wikimedia.org/w/index.php?title=Submissions%2FResourceLoader_2%3A_The_Future_of_Gadgets&diff=30139&oldid=25927 [23:11:53] RoanKattouw_away: any comment on https://gerrit.wikimedia.org/r/#change,6463 ? [23:11:56] RoanKattouw_away: ok [23:12:29] TimStarling: Looking [23:12:45] "Don't even THINK about putting set -e here" [23:12:45] ^demon: just wondering "use https" in what sense, if I may ask? [23:12:47] lol [23:12:52] ignore whitespace to see the changes [23:13:29] -# Don't even THINK about putting set -e here [23:13:33] +» set -e [23:13:36] I saw ;) [23:13:39] I know the subshell makes it okay, but still, I loled [23:13:53] ^demon: thanks [23:13:55] * TimStarling is not good at following orders [23:14:31] * AaronSchulz thinks tim is unfit for the army [23:14:48] I tested it using a copy in my home directory on fenari, I ran it a couple of times [23:14:53] Hmm, -F5 [23:15:10] That would be good for large syncs (like new code trees) yes [23:15:17] that's what it needed when we were pushing out a whole new version, a couple of versions ago [23:15:38] <^demon> Krinkle: Submodules in wmf branch use ssh right now. There's no need, since we're not pushing back to the upstream projects. [23:15:47] Well I think it might be unreasonably slow for smaller scaps [23:15:58] ^demon: ah, so changing it to an anonymous clone then, right? [23:16:07] So maybe it should default to 30 and accept a flag that lowers it? [23:16:19] <^demon> Well maybe not of core, but the submodules can be anon. [23:16:22] I thought maybe you were talking about changing canonical server to https or something :P - context is everything [23:16:27] let me time it [23:16:52] I assume there's no pending changes right now, so this will be a baseline minimum time [23:17:03] <^demon> Ryan_Lane: Submodules are weird at first, but if you spend a couple of hours playing with them and making mistakes you end up with a pretty good grasp on them. Like most things in git they make sense once you play with it enough. [23:17:10] RoanKattouw_away: how about a low default and a flag to speed things up? [23:17:24] yeah. I have it down now [23:17:29] btw ~mwbot on toolserver is now using protocol-relative urls :) [23:17:37] TimStarling: Actually there are some PageTriage changes but the ext is only enabled on testwiki so it's safe [23:17:52] not a huge amount of data to transfer though? [23:18:17] Reedy: Do you know if mwbot stuff is in some kind of repo? It think it was originally in a repo once, but got developed out of repo further, correct? [23:18:23] * TimStarling tries to think of a command that will attach timestamps to stdout output [23:18:30] AaronSchulz he is fit for the army. He just needs to be the comander in chief [23:18:32] Nah small changse [23:18:37] there's always strace -tt I guess [23:18:44] TimStarling: rsync -v ? [23:18:50] awjr: https://www.mediawiki.org/wiki/User:Chughakshay16/ConventionExtension [23:18:59] I'd have to change the remote scripts to do that [23:19:10] oh yeah, he is one of the dudes i was working with in pune [23:19:14] i'll get in touch with him [23:19:17] awjr: cool! [23:19:29] awjr: and Jure (freakolowsky) is his mentor. [23:21:03] awjr: and I'm right in inferring that the forms stuff would be useful to him, right? [23:21:14] hi kaldari! is tomorrow a heavy GSoC/UploadWizard day for you? [23:21:37] not heavy, but not light :) [23:21:43] @dump [23:21:43] Krinkle: probably... wouldn't suprise me [23:21:44] A dump is now available at http://toolserver.org/~mwbot/botbrain.html [23:21:45] A dump is now available at http://toolserver.org/~mwbot/botbrain.html [23:22:08] kaldari: you saw my note about what I'd like the students to have done by 21 May? [23:22:19] yes [23:22:27] I'm in good shape for that [23:22:32] cool, great, thanks kaldari! [23:25:39] sumanah: most likely [23:27:27] bsitu: can I ask for you to do a little code review in your 20% time tomorrow on this patch? https://bugzilla.wikimedia.org/show_bug.cgi?id=14901 [23:28:01] sumanah: yes [23:28:55] bsitu: ok, in that case, assigning to you. This is a problem that is affecting Wikimedians right now and it would be best if it could be fixed fairly soon. robla set the priority at normal, though, so it's not the most urgent thing ever. [23:30:17] <^demon> Ryan_Lane: Ok fixed in make-wmf-branch for the next release. [23:30:18] sweet. 1.20 works perfectly fine with labsconsole testwiki [23:30:51] hume: No entry for terminal type "unknown"; [23:30:51] hume: using dumb terminal settings. [23:31:00] the positive of using basically only extensions we use in production :) [23:31:11] yesterday I worked out where that message comes from [23:31:28] anyone interested? [23:31:47] <^demon> I had the same error before, I'm curious if you had the same cause & solution. [23:32:03] it's from libedit [23:32:16] * AaronSchulz is interested [23:32:17] PHP's readline extension initialises libedit on MINIT [23:32:50] libedit calls tgetent() which apparently returns zero for "unknown" even when /usr/share/terminfo/u/unknown exists [23:33:10] however it doesn't return zero for "dumb" which is the actual default used for a normal ssh session [23:33:37] when you log in via ssh, TERM is in fact unset, and sudo has a very helpful hard-coded fallback [23:33:48] if there is no TERM, it sets it to "unknown" [23:34:04] having it unset would just make it use "dumb" and there would be no error [23:34:36] so basically the error comes from cooperation between ssh, sudo, php and libedit and could be fixed in any of those places [23:35:27] but I think putting TERM=dumb in scap-2 is probably the most practical fix [23:35:27] <^demon> Which one will fix it? [23:36:46] also compiling php with readline instead of libedit would fix it [23:37:01] because readline isn't so retarded