[00:09:38] <greg-g>	 marktraceur: Elsie: please file a bug so this issue can be resolved before the next deployment. Importance->High i'd say.
[00:09:44] <greg-g>	 don't want to lose it
[00:10:20] <rillke>	 Api is down.
[00:10:22] <rillke>	 https://meta.wikimedia.org/w/api.php?format=jsonfm&action=query&meta=globaluserinfo&guiprop=merged&guiuser=Rillke
[00:10:43] <Technical_13>	 Again?
[00:10:52] <greg-g>	 again?
[00:10:52] <Elsie>	 en.wikipedia.org seems fine.
[00:10:53] <rillke>	 Wikimedia Foundation Error Request: GET  http://meta.wikimedia.org/w/api.php?format=jsonfm&action=query&meta=globaluserinfo&guiprop=merged&guiuser=Rillke,  from 208.80.152.16 via sq62.wikimedia.org (squid/2.7.STABLE9) to  ()
[00:10:53] <rillke>	 Error: ERR_CANNOT_FORWARD, errno (11) Resource temporarily unavailable at Fri, 30 Aug 2013 00:10:26 GMT 
[00:10:59] <greg-g>	 ugh, same error
[00:10:59] <Elsie>	 Technical_13: What do you mean by again?
[00:11:06] <Elsie>	 It loaded for me.
[00:11:10] <Elsie>	 Took a bit, though.
[00:11:13] <greg-g>	 Elsie: it happened today around.... 3pm pacific?
[00:11:18] <Technical_13>	 Api was down earlier too.
[00:11:21] <jem->	 Problemas in es.wiki also
[00:11:21] <greg-g>	 Elsie: the issue was with esams
[00:11:26] <Elsie>	 What'd you do to the API?
[00:11:33] <greg-g>	 so, not an issue for us 'mericans
[00:11:39] <Elsie>	 Hmm.
[00:11:45] <Technical_13>	 addshore broke it.
[00:11:57] <Technical_13>	 At least that is my theory...
[00:12:00] <jem->	 And in every wiki I'm testing of
[00:12:15] <greg-g>	 jem-: you're located in Europe, right?
[00:12:41] <greg-g>	 rillke: where are you physically located? which continent?
[00:12:42] <rillke>	 Yes, I am.
[00:12:45] * greg-g  nods

[00:13:01] * greg-g  goes to read backscroll from when this happened earlier today

[00:13:02] <rillke>	 Can I somehow connect to the U.S. Servers?
[00:13:10] <Elsie>	 I don't see anything in the SAL.
[00:15:57] <Elsie>	 commons seems slow for me.
[00:19:50] <jem->	 greg-g: Yes, in Spain more precisely
[00:20:04] <Ryan_Lane>	 rillke: working now?
[00:20:04] <Ryan_Lane>	 jem-: ^^ ?
[00:20:22] <jem->	 Not yet
[00:20:46] <jem->	 (I'm checking just the API with my bot)
[00:21:01] <rillke>	 No.
[00:21:08] <Ryan_Lane>	 hm, well I switched the service to a working datacenter
[00:21:17] <rillke>	 Error: ERR_CANNOT_FORWARD, errno (11) Resource temporarily unavailable at Fri, 30 Aug 2013 00:20:56 GMT 
[00:22:06] <jem->	 I must go to bed... good luck with it :)
[00:22:08] <rillke>	 Do I have to flush my DNS cache?
[00:26:15] <rillke>	 https://commons.wikimedia.org/w/api.php?action=tokens&format=jsonfm also fails
[00:26:26] <rillke>	 this means UploadWizard fails
[00:28:14] <Ryan_Lane>	 rillke: no, it seems my change didn't switch things
[00:31:37] <rillke>	 It's back.
[00:58:08] <helderwiki>	 Hi! Was there any recent config change which might have caused the bold items from my https://pt.wikipedia.org/wiki/Special:Watchlist not to be bold anymore?
[00:59:17] <Reedy>	 How recent? 6-7 hours ago?
[02:40:52] <kaldari>	 Looks like the HTTPS switch broke Flickr importing on Commons. Anyone want to +2 the fix: https://gerrit.wikimedia.org/r/#/c/81890/
[02:47:48] <Krenair>	 kaldari, default UploadWizard config should be modified as well
[02:48:03] <kaldari>	 true, i'll do that now
[04:10:31] <aude>	 wtf, i am logged out and seeing commons in the dyslexic font
[05:58:14] <Nemo_bis>	 oh, overzealous magic linking https://bugzilla.wikimedia.org/show_bug.cgi?id=37583#c3
[06:00:07] <Nemo_bis>	 ori-l: could you check this in the logs for me? https://bugzilla.wikimedia.org/show_bug.cgi?id=28827#c19
[06:00:38] <ori-l>	 sure
[06:01:53] <ori-l>	 it happened again today
[06:02:03] <ori-l>	 [29-Aug-2013 15:45:50] Catchable fatal error: Argument 1 passed to EditPage::__construct() must be an instance of Article, null given, called in /usr/local/apache/common-local/php-1.22wmf14/extensions/LiquidThreads/classes/View.php on line 734 and defined at /usr/local/apache/common-local/php-1.22wmf14/includes/EditPage.php on line 272
[06:02:24] <ori-l>	 let's see what lqt is trying to do there
[06:03:38] <greg-g>	 ori-l: you should go to sleep soon
[06:03:57] <ori-l>	 probably
[06:04:45] <Reedy>	 lol
[06:05:05] <greg-g>	 Reedy: I don't even know what to say to you ;)
[06:05:23] <greg-g>	 Reedy: how long have you been up?
[06:06:30] <Reedy>	 16 hours or so I think
[06:07:37] <greg-g>	 golly gee
[06:07:43] <greg-g>	 or something
[06:08:16] <Reedy>	 I went to bed 3 or 4 hours ago but apparently couldn't sleep
[06:08:40] <greg-g>	 ok, time for me to go. 4 things checked off my "moving out of this crazy house" todo list
[06:08:47] <greg-g>	 Reedy: ugh
[06:08:52] <greg-g>	 good luck for tomorrow/today
[06:10:07] <ori-l>	 LQT has methods like 'assertSingularity'
[06:10:18] <ori-l>	 I'm not making that up
[06:11:09] <ori-l>	 good night greg-g
[06:11:31] <Nemo_bis>	 hmm ori-l, I've pasted your line in the bug but now I wonder if it actually is the same thing
[06:12:03] <ori-l>	 dunno yet
[06:12:23] <Nemo_bis>	 LQT has plenty of bugs with X must be an instance of Y fatals
[06:15:03] <ori-l>	 this must be an instance of one of those
[06:25:59] <ori-l>	 Nemo_bis: sorry, I'm running out of steam tracing this down
[06:26:09] <ori-l>	 i got to a comment that reads: 		// Yuck. 
[06:26:36] <ori-l>	 i'd be happy to keep looking tomorrow, feel free to ping me about it
[06:36:27] <Nemo_bis>	 ori-l: haha, LQT does that
[06:38:18] <Nemo_bis>	 it seems Krenair has a special mithridatisation against LQT madness, we should isolate his anti-toxin and distribute the vaccine
[11:28:38] <cortexA9>	 hello, i wanna help wikimedia, how can i contribute ?
[11:28:39] <cortexA9>	 i mean in the tech field
[11:36:30] <MatmaRex>	 cortexA9: https://www.mediawiki.org/wiki/Special:MyLanguage/How_to_contribute
[11:37:00] <cortexA9>	 MatmaRex: thanks
[11:37:30] <MatmaRex>	 cortexA9: the good stuff for developers is at https://www.mediawiki.org/wiki/Developer_hub :)
[11:40:28] <cortexA9>	 MatmaRex: it's not possible to contribute in the sysadmin operations?
[11:41:49] <MatmaRex>	 cortexA9: i don't really know, sorry
[11:42:09] <cortexA9>	 ok no problem thanks MatmaRex
[11:42:24] <MatmaRex>	 cortexA9: but wikimedia's operations stuff is usually handled by wmf's operations team :)
[11:42:49] <MatmaRex>	 cortexA9: you can help with e.g. coding up configuration changes projects ask for, but that's usually not too exciting
[11:43:37] <MatmaRex>	 cortexA9: their channel is #wikimedia-operations, feel free to hang out there :)
[11:44:07] <cortexA9>	 MatmaRex: good, thanks :)
[13:15:19] <jeblad>	 Anyone that can explain why the the search results in the drop down menu suddenly has a very low hit rate?
[13:17:08] <p858snake|l>	 jeblad: where? mw wiki?
[13:17:47] <jeblad>	 Its at nowiki, but I've heard people say it is also at other wikis
[13:18:19] <jeblad>	 I have only observed it on nowiki and it seems like obvious articles are left out
[13:19:29] <gry>	 central login has been logging me in to wikis without actually properly applying user settings such as skin
[13:19:39] <gry>	 i feel it's a new thing, started happening today
[13:20:03] <gry>	 with popup, similar to the 'edit was saved' one (never saw these before either)
[13:20:15] <p858snake|l>	 you need to refresh for user perfernces such as skins, you have always needed to do that
[13:20:22] <p858snake|l>	 gry: that message has been around for ages
[13:20:34] <p858snake|l>	 unless you had js to disable it
[13:20:46] <gry>	 edit message yes, 'central bla bla logged you in' is new
[13:21:35] <gry>	 i mean edit message is old, but the central auth one is new, i know these popups but they appeared from central auth only today
[13:21:39] <jeblad>	 I think the skin is failing more often now, I been asked about it several times lately
[13:22:27] <gry>	 is there a way to disable these popups without js tricks btw ? i would prefer messages similar to 'you have new talk messages' which are static and dismissable
[13:22:45] <gry>	 without having to re-do js stuff at each new computer
[13:23:29] <gry>	 them moving around is more or less ok but them /popping up/ disturbs me
[13:53:19] <Clinteger>	 okay, I am still having problems where rather often page load takes a long time because requests to mediawiki's resourceloader does not complete. is this a known problem? 
[15:01:49] <svick>	 parent5446: hello
[15:02:44] <parent5446>	 Hey
[15:03:14] <parent5446>	 apergos?
[15:04:05] <svick>	 apergos is on vacation today (and i think also on monday)
[15:04:47] <parent5446>	 Ah OK, well for some reason he's still on IRC
[15:05:25] <svick>	 yeah
[15:06:13] <svick>	 so, today i'm working on using zdelta (to compare it with other options)
[15:07:53] <parent5446>	 OK. How many algorithms are you comparing in total again?
[15:09:08] <svick>	 well, according to that paper, zdelta is best for delta compression, so i'm not planning on trying another delta compression algorithm
[15:09:25] <svick>	 but i want to compare it with using LZMA on a group of revisions
[15:09:43] <parent5446>	 OK. Makes sense since zdelta is L77-based.
[15:11:07] <svick>	 and in both cases, i want to try some options, like the size of a group or what happens if i don't make one huge delta chain for all revisions of a page, but a separate chain for each n revisions (so that loading the last revisions doesn't require all the previous ones)
[15:14:03] <svick>	 and first results look somewhat promising: the tenwiki dump i'm testing it on has 100 MB in umcompressed XML, 2 MB as xml.7z, if i remember correctly, ~30 MB each revision separatelly compressed using LZMA, and ~5 MB for zdelta
[15:14:35] <svick>	 that's still more than 2x the size of xml.7z, but i think it could work
[15:14:57] <parent5446>	 Hmm, yeah it's still bigger, but still has a good compression ratio.
[15:16:19] <svick>	 yeah
[15:17:11] <parent5446>	 OK, well I don't have anything else. Hopefully the test results stay the same or get better.
[15:17:27] <svick>	 yeah, we'll see about the other options
[15:17:48] <svick>	 i don't have anything either; see you monday
[15:17:55] <parent5446>	 See you monday
[16:04:22] <AdamCuerden>	 https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2013-08-28/Technology_report <- If anyone sees anything that's been left out of this, please let me know
[16:08:18] <AdamCuerden>	 I'm holding back the new VE buttons until they get a bit nearer en-wiki, since "Faster loading" is a pretty good VE lead for this week. 
[16:08:48] <AdamCuerden>	 Galleries get a big writeup as they've just reached en-wiki.
[16:08:58] <AdamCuerden>	 And, well, they're kind of gamechangers
[16:24:51] <bawolff>	 AdamCuerden: In regards to your post to wikitech-ambassadors about if you missed anything - The campaign namespace already exists: example: https://commons.wikimedia.org/wiki/Campaign:wlm-nl YuviPanda is currently working on making a less technical pretty display for them like http://blue-dragon.wmflabs.org/wiki/Campaign:show-off-campaigns
[16:28:00] <AdamCuerden>	 Right
[16:28:14] <AdamCuerden>	 Let's see...
[16:28:40] <AdamCuerden>	 I think that got edited down in Kirill's copyedit
[16:30:28] <AdamCuerden>	 * '''Campaign infrastructure improvements planned for Commons''': [[:mw:User:Yuvipanda/Campaigns_namespace_proposal|Improvements to the "campaigns" namespace]] have been proposed for Commons. The proposed infrastructure will allow programs like Wikipedia Loves Monuments to easily preformat file information and document uploads. Although a [[:commons:Campaign:wlm-nl|ba
[16:30:29] <AdamCuerden>	 sic form of the idea]] exists already, this should vastly improve efficiency.
[16:30:35] <AdamCuerden>	 Does that sound good?
[16:31:01] * AdamCuerden  appends "and useability

[16:41:16] <bawolff>	 sounds good to me
[16:42:17] <marktraceur>	 AdamCuerden: s/useability/usability/
[18:11:30] <AdamCuerden>	 MArk: Technically, that's a BE / AE issue. 
[18:11:53] <marktraceur>	 AdamCuerden: Me?
[18:12:57] <AdamCuerden>	 Aye. Isn't "Useability" the British spelling? 
[18:14:05] <mutante>	 wiktionary calls it simply alternative spelling but doesn't claim British/American
[18:14:10] <marktraceur>	 I um
[18:14:14] <AdamCuerden>	 Hmm.
[18:14:15] <marktraceur>	 I didn't think that was true
[18:14:27] <AdamCuerden>	 Eh, well. Usability looks weird.
[18:14:27] <marktraceur>	 James_F: "Useability" - en_GB or just en?
[18:15:07] <domas>	 "variant spelling"
[18:15:15] <domas>	 according to OED
[18:15:15] * marktraceur  thinks there's a used car marketing campaign in "USA-bility", maybe for President's Day or something

[18:15:17] <greg-g>	 all you variants
[18:35:22] <AdamCuerden>	 Eh, well. Acceptable variant is acceptable variant  =)
[18:40:19] <mutante>	 http://en.wiktionary.org/wiki/variant#Alternative_forms
[18:40:27] <marktraceur>	 Hah
[18:40:44] <marktraceur>	 Yo dawg, I herd you like varients.
[18:41:31] <mutante>	 variaunt vagrant = future Ubuntu release?
[18:41:52] <mutante>	 https://github.com/DataDog
[18:42:16] <mutante>	 <-- open source stuff of https://www.datadoghq.com/
[18:42:50] <greg-g>	 mutante: IF it gets that far :)
[18:43:20] <mutante>	 hah, yea
[18:56:34] <AdamCuerden>	 What is Ubuntu up to? U or so?
[18:56:45] <marktraceur>	 Q, methinks
[18:56:55] <marktraceur>	 October will be R IIRC
[18:57:35] <marktraceur>	 Oh, maybe April was R
[18:58:09] <marktraceur>	 "Raring Ringtail"
[18:58:13] <marktraceur>	 Next up "Saucy Salamander"
[18:58:22] <YuviPanda>	 better than sassy salamander?
[18:58:47] <brion>	 Sadistic Sasquatch
[18:59:22] <marktraceur>	 Wonder what moneybags will use for T
[18:59:39] <marktraceur>	 Oh, it'll be an LTS
[18:59:46] <AdamCuerden>	 Terrifying Tapir
[18:59:47] <marktraceur>	 The adjective had better be "Trusty"
[19:00:01] <marktraceur>	 Trusty Turtle
[19:00:12] <YuviPanda>	 Trusty Tortoise, you mean
[19:00:13] <YuviPanda>	 :P
[19:00:17] <marktraceur>	 Ooh, better.
[19:00:25] <marktraceur>	 Trusty Tortoise, you heard it here first
[19:00:49] <marktraceur>	 Though they might go with Tapir
[19:01:00] <marktraceur>	 Them liking their exotic animals so much
[19:01:32] <Ryan_Lane>	 Terrifying Tarantula
[19:01:34] <YuviPanda>	 Treacherous Tarantula?
[19:01:37] <YuviPanda>	 gah, beat me
[19:01:39] <Ryan_Lane>	 ;)
[19:01:44] <YuviPanda>	 Terrible Tarantula, perhaps
[19:01:46] <Platonides>	 it would be a problem for them to decide between Tortoise and Turtle :P
[19:01:49] * YuviPanda  likes spiders, they're cute.

[19:02:23] <Ryan_Lane>	 Platonides: if it was the wikimedia community deciding that, we'd have 80 RFCs and we'd have massive fights after one was actually decided upon
[19:02:40] <Ryan_Lane>	 then a staff member would mispell it
[19:04:08] <YuviPanda>	 and then there'd be 20 threads on different VPTs asking for all the staff to be fired
[19:04:16] <YuviPanda>	 VPs, not VPTs
[19:04:42] <mutante>	 and then of course.. the  Unbelievable Unicorn
[19:04:55] <Ryan_Lane>	 mutante: that would be a great name
[19:04:58] <Ryan_Lane>	 too bad that's not an LTS
[19:05:16] <mutante>	 hah, yea:p
[19:05:43] <marktraceur>	 Ryan_Lane: October 2014 is obviously when they start rolling releases - Unstable Unicorn will be that branch, and they'll start releasing a separate line for stable use
[19:05:45] <YuviPanda>	 Unfindable Unicorn?
[19:06:03] <YuviPanda>	 they'll probably start using absolutely random chinese words :P
[19:06:05] <mutante>	 if it was LTS it should be 'unbreakable'
[19:06:38] <brion>	 emoji
[19:06:45] <YuviPanda>	 'pile of shit'
[19:07:01] <mutante>	 Promising Panda
[19:07:16] <Ryan_Lane>	 marktraceur: eh? I thought the code name was directly associated with the date release?
[19:07:17] <YuviPanda>	 Panda is clearly not obscure enough for them
[19:07:25] <brion>	 💩 
[19:07:40] <YuviPanda>	 brion: hehe, was too lazy to search
[19:07:41] <marktraceur>	 Ryan_Lane: 14.10 will be the "U" release
[19:07:51] <marktraceur>	 2014, 14, October, 10
[19:07:51] <brion>	 Utopian Unicorn
[19:07:58] <Ryan_Lane>	 yeah, that's not an LTS
[19:08:05] <marktraceur>	 Right, 14.04 is
[19:08:06] <Ryan_Lane>	 oh
[19:08:07] <Ryan_Lane>	 wait
[19:08:08] <Ryan_Lane>	 it is
[19:08:11] <marktraceur>	 Which will be a "T" release
[19:08:25] <Ryan_Lane>	 I keep forgetting we're in 2013
[19:08:32] <brion>	 Tempting Trilobite
[19:08:47] <marktraceur>	 Heh
[19:08:55] <mutante>	 i used to have a trilobite
[19:08:58] <marktraceur>	 Even number .04 releases are LTS
[19:09:16] <YuviPanda>	 Troubled Tribbles
[19:09:23] <brion>	 :D
[19:09:31] <YuviPanda>	 ST + Ubuntu!
[19:09:36] <marktraceur>	 YuviPanda: Oh no, they're totally going to do that
[19:09:38] <marktraceur>	 Damn it
[19:09:47] <YuviPanda>	 I only wish.
[19:09:55] <marktraceur>	 I was so excited about "Trusty" and the potential of tapirs.
[19:10:11] <YuviPanda>	 Tribbles got that beat
[19:10:18] <mutante>	 so if we come up with release names for labs or something ..Worrisome Wikipedian
[19:10:43] <marktraceur>	 Totally
[19:11:04] <marktraceur>	 mutante: Dapper Dramah
[19:11:14] <marktraceur>	 Or Deletionist
[19:11:34] <YuviPanda>	 Interpid Inclusionist?
[19:12:13] <James_F>	 marktraceur: Useability. "Usability" Would canonically be pronounced US-AH-BIL-IT-IE, not EUSE-AH-BIL-IT-IE.
[19:12:37] <marktraceur>	 James_F: No it wouldn't - the "a" after the "s" means "u" is pronounced "you"
[19:13:16] * marktraceur  goes to lunch to debate this further

[19:27:11] <Choock>	 Hi all
[19:45:08] <kaldari>	 robla: looks like we're converting the page into a DOMDocument and then using Xpath to remove the stuff we don't want. Obviously not a very performant solution.
[19:47:35] <kaldari>	 robla: see MobileFrontend/includes/formatters/HtmlFormatter.php
[20:13:10] <MaxSem>	 kaldari, could've been worse: https://graphite.wikimedia.org/dashboard/temporary-29 :)
[20:17:59] <kaldari>	 MaxSem: what are these graph lines showing?
[20:19:54] <MaxSem>	 total transformation time
[20:20:06] <kaldari>	 what's the difference between the 2 lines?
[20:20:21] <MaxSem>	 avg vs 99th percentile
[20:20:23] <kaldari>	 ah
[20:20:39] <kaldari>	 and this is in milliseconds?
[20:20:48] <kaldari>	 or seconds?
[20:20:53] <MaxSem>	 ms
[20:20:54] <kaldari>	 :)
[20:21:07] <MaxSem>	 seconds wouldv'e been a disaster:)
[20:22:44] <kaldari>	 that's not as bad as I would expect, especially for the 99th %.
[20:25:46] <robla>	 MaxSem: fwiw, Chad is looking into an HTML parser that Google wrote as something to bolt in as a C extension
[20:26:24] <brion>	 their new pure-C HTML5 parser? that has a certain appeal yes
[20:26:50] <brion>	 iirc speed wasn't a goal but i haven't compared it to libxml2 for speed
[20:26:51] <robla>	 brion: I think so.  all I know is what I learned in our 2 min conversation on the subject
[20:26:55] <brion>	 for compatibility, it should be way better
[20:27:04] <brion>	 libxml2 occasionally really fucks up odd-looking html
[20:27:59] <MaxSem>	 robla, for which purpose?
[20:28:23] <robla>	 MaxSem: they need to strip out some navigation stuff prior to indexing for search
[20:28:34] <robla>	 manybubbles: you may be interested in this conversation ^^
[20:28:52] <MaxSem>	 heh, I had entertained myself with similar stuff too:)
[20:29:34] <manybubbles>	 hey, yeah!
[20:29:37] * robla  wonders about this: https://code.google.com/p/streamhtmlparser/

[20:30:18] <manybubbles>	 so right now I've got a patch set that uses https://github.com/Masterminds/html5-php and it is slow.  We're going to sit on it for a bit.
[20:30:19] <robla>	 meh, dead project, nevermind
[20:30:24] <ori-l>	 what are you trying to do?
[20:30:43] <robla>	 strip out some navigation stuff prior to indexing for search
[20:31:30] <robla>	 I'm not sure if there's an equivalent of Sax for HTML (or if Sax is more suitable for it these days)
[20:32:09] <manybubbles>	 ori-l: pretty much.  What I wanted was to be able to $('.whatever').remove() from php but what I got was xpath
[20:32:12] <manybubbles>	 which isn't too bad.
[20:32:34] <ori-l>	 well, i know that using string processing methods to handle HTML is a way to get hospitalized in an asylum, but
[20:32:51] <ori-l>	 mediawiki html is uniform in some key respects
[20:32:54] <MaxSem>	 manybubbles, is that parser faster than libxml? though yeah, I guess any random parser you find on the internetz is way less buggy than it:}
[20:33:19] <ori-l>	 i wonder if you could just grep for <div id="bodyContent">
[20:33:24] <manybubbles>	 in any case, it looks like our options for actually parsing html5 in php are that html5-php which is slow-ish and gumbo-parser which doesn't support php yet so isn't really an option now.
[20:33:26] <ori-l>	 i guess you need to figure out the matching </div> tag
[20:33:35] <YuviPanda>	 ori-l: well, http://stackoverflow.com/a/1732454/17865
[20:33:46] <ori-l>	 yes, i know :)
[20:34:01] <YuviPanda>	 no NO NOO̼O​O NΘ stop the an​*̶͑̾̾​̅ͫ͏̙̤g͇̫͛͆̾ͫ̑͆l͖͉̗̩̳̟̍ͫͥͨ
[20:34:26] <manybubbles>	 so you can do some stuff with a regex - it is pretty fragile and you end up relying on the whitespace behavior of our html renderer
[20:34:30] <manybubbles>	 which is not a good idea
[20:34:45] <ori-l>	 we could also patch mediawiki to add <!-- content ends here --> comment if there isn't a suitably unique string that demarcates the end of page-specific content
[20:34:51] <manybubbles>	 libxml doesn't support html5.  it just blows up
[20:35:15] <manybubbles>	 it is worse than that though - we need to remove some content from the page because it shouldn't be searchable.
[20:35:18] <bawolff>	 does mediawiki output html5 specific things?
[20:35:24] <MaxSem>	 wait, what do you grab, manybubbles?
[20:35:24] <manybubbles>	 stuff like the contents of <audio> and <video> tags
[20:35:35] <ori-l>	 oy vey
[20:35:44] <manybubbles>	 the table of contents too
[20:36:00] <bawolff>	 what sort of html5 thingies cause it to blow up?
[20:36:03] <manybubbles>	 basically it is just the table of contents and the contents of <audio> and <video> tags now
[20:36:21] <ori-l>	 is that likely to change a lot?
[20:36:33] <ori-l>	 i think that's still just barely under the maximum threshold for raw string manipulation
[20:36:39] <manybubbles>	 bawolf: <audio> and <video>.  it works when you parse as xml but our pages aren't valid xml either.
[20:36:41] <ori-l>	 if you expect to need to filter out many more things then yes, you need to parse
[20:36:57] <manybubbles>	 ori-l: maybe
[20:36:58] <robla>	 bawolff:  I believe this is valid html5: <b><i>whee!</b></i>
[20:37:13] <MaxSem>	 manybubbles, TOC is available separately from ParserOutput
[20:37:18] <manybubbles>	 robla: html5 is really really permissive
[20:37:43] <manybubbles>	 MaxSem: unless I'm missing something it isn't available in quite the way I need.
[20:38:12] <robla>	 manybubbles: yeah, this is probably valid HTML too:  <a;lsdjayvchxcvkjasdf>&urg!>;<andanotherthing/yadda>/>
[20:38:17] <MaxSem>	 there's getTOCHTML()
[20:38:34] <manybubbles>	 MaxSem: Yeah but there is sausage making that gets in the way
[20:38:34] <ori-l>	 it's not an html output problem though, it's a mediawiki html output problem, which is a different beast
[20:38:39] <MaxSem>	 and getSections(0 that contains a lot of metadata
[20:39:16] <manybubbles>	 MaxSem: the problem with getTOCHTML is that that html isn't the same html that gets put in the text - it gets munged first.
[20:39:29] <manybubbles>	 MaxSem: so you can't just find it and replace it with empty string
[20:39:34] <manybubbles>	 which made me sad
[20:39:43] <greg-g>	 robla: mind running the roadmap update tool again? I haven't had time to futz with it yet
[20:40:05] <greg-g>	 "<ctrl>+r roadmap" should probaby do it ;)
[20:40:26] <MaxSem>	 I'm actually working on a way to return HTML w/o TOC  - https://gerrit.wikimedia.org/r/80578
[20:40:30] <MaxSem>	 manybubbles, ^^
[20:40:34] <ori-l>	 how about everything from: "<h1 id="firstHeading" to "<!-- NewPP limit" ?
[20:40:46] <manybubbles>	 ori-l: yeah, because the problem is limited to mediawiki I can be more confident that string manipulation won't totally fail.  it doesn't make me feel good though.
[20:41:02] <ori-l>	 manybubbles: i know exactly what you mean
[20:41:23] <ori-l>	 like if you give a presentation at a conference about this you'd have to cough loudly and skip this part :P
[20:41:38] <ori-l>	 bbl
[20:42:13] <robla>	 greg-g: something is messed up about my local install...sorry
[20:43:23] <YuviPanda>	 (vagrant!)
[20:44:08] <greg-g>	 robla: blugh. Sumanah promised in her email to fix this too, right? :)
[20:44:15] <greg-g>	 s/too/tool/
[20:44:28] <robla>	 heh
[21:18:18] <Platonides>	 manybubbles, you can request the article without the TOC
[21:19:00] <manybubbles>	 Platonides: this?  https://gerrit.wikimedia.org/r/#/c/80578/
[21:44:35] <manybubbles>	 Nemo_bis: would you like me to add as a reviewer on more CirrusSearch stuff?