[08:23:41] HAHA finally after months of waiting my change in puppet for bots project was merged, just when the bots project is getting deleted... this kind of proves how unusable the current gerrit + puppet is for people who don't have +2 [08:24:46] I need to store a link for that for people who keep asking me why I don't use puppet [08:25:54] petan: why isn't there a seperate labs/puppet repository? [08:25:55] er [08:25:59] tools labs-puppet* [08:31:59] I have no idea I don't do gerrit, you need to ask demon [08:43:06] do we have German-speaking folks here? [08:45:27] gry: Steinsplitter [08:45:39] gry: or me [08:46:20] [22:14:41] would someone with German fluent knowledge please read https://meta.wikimedia.org/wiki/Grants:IEG/LaTeX_Export and its talk page, and ask the guy what he actually needs? a sysadmin? I got the impression everything can be done for free and requested mailing lists)> [08:46:44] (it puzzles me as to why he doesn't understand that labs are already maintained and free) [08:48:42] gry: sounds like "i wrote cool software, it works on my machine, now go use it!" [08:48:52] i don't see any german there [08:49:16] As you may note from the talk page, it is a folk who can speak English, but he does not understand me. [08:49:31] I was thinking probably a language barrier. [08:51:41] I don't think it's a language problem. I think it's a problem of understanding organisational structures and processes. [08:52:03] Essentially, he is asking for an extension to be instaleld and enabled (plus a backend service/software package). [08:52:12] I can't see how that fits into the grant program, though [08:52:16] [08:52:20] what is this place? [08:52:49] bugzilla is the right place for requesting this [08:53:03] after getting community support on the village pump ertc [08:53:21] ok, so I've added a message on that talk page; will see how it goes [08:53:25] thanks [08:53:47] since there is no work to be done on their side, nor do they need to invest anything, i don't see how a grant would help. [08:54:19] yeah, I don't care of the grant; more of reaching understanding somehow [08:54:47] but you'll need to explain to him why his request is in the wrong place :) [09:04:23] $1K/server, really? [09:04:50] Target audience: Everyone who wants to download high quality PDF Version of Wiki Pages [09:05:10] check, we already have PDF export [09:05:37] which, I know, sucks major time:P [09:13:26] well [09:14:28] http://pastebin.mozilla.org/2923059 is this description accurate? [09:15:26] what is mostly done, Wiki2Latex? [09:16:09] gry, ^^ [09:16:19] yes, he says he already has binaries, so [09:17:20] see anything at https://www.mediawiki.org/wiki/Extension:Wiki2LaTeX ?:P [09:18:20] oh, sigh. I'll now have to find whether it is the one he is writing, or not [09:18:21] this whole bloody thing needs to be rewritten from scratch if someone wants to see it anywhere near WMF [09:18:48] even labs shouldn't host this kind of crap [09:19:19] can't we use parsoid to generate latex ? :D [09:19:24] if it requires running in a VM as he writes, it's not suitable for WMF, period [09:19:44] he said it can be run on the cloud-ish thing on debian, which is what labs has, I think [09:19:49] hashar, yeah https://en.wikipedia.org/w/index.php?title=Henry_Cavill&diff=570344047&oldid=570306976 [09:19:52] anyway I'll add a note about that [09:20:12] MaxSem: :))) [09:23:08] MaxSem: I think the one you found has a similar name, but is not it. [09:24:00] gry, someone enquired me yesterday about that particular extension yesterday, likely it's related [09:24:18] speaking of the devil - hashar, we're running tests in jenkins on sqlite, right? [09:24:24] the one you linked is here (https://code.google.com/p/wiki2latex/) and the one the guy develops is here http://sourceforge.net/projects/wb2pdf/files/ [09:24:28] MaxSem: right [09:24:40] MaxSem: haven't managed to start working on a mysql wrapper [09:24:59] hashar, cause I don't seem to be able to run tests on MySQL anymore:P [09:26:56] :( [10:07:13] kowiki database is not available on toollabs? [10:07:23] MariaDB [kowiki_p]> select count(*) from user; [10:07:26] ERROR 1356 (HY000): View 'kowiki_p.user' references invalid table(s) or column(s) or function(s) or definer/invoker of view lack rights to use them [10:07:57] (03PS2) 10Hashar: pass pep8 v1.4.6 [labs/nagios-builder] - 10https://gerrit.wikimedia.org/r/80963 [10:08:04] (03PS3) 10Hashar: pass pep8 v1.4.6 [labs/nagios-builder] - 10https://gerrit.wikimedia.org/r/80963 [10:12:00] let me know whom I need to contact, pls [10:12:55] I am trying to access on tools-dev and I succeeded to query on enwiki. [10:13:37] I assume it does not clone for some projects. [10:14:37] Maybe I can request for additional cloning, but I don't know how to ask. [10:35:18] Coren, ^^^ [12:33:14] ryuch_: In theory everything is replicated. What project are you trying to access? [12:56:21] hi, Coren. Korean Wikipedia, kowiki_p [12:58:13] ryuch_: Ah. It's another one of those "there are two databases with that name and the default is the wrong one". Fixed, you should be able to connect to it now. [13:00:07] thanks i will try [13:01:48] thanks, it works. [13:01:55] great! [13:03:01] maybe i am the firt who acceses kowiki on labs. [13:08:20] Hi all! Do you know if Vera is here on IRC? She was in the workshop at Wikimania and was one of the people with a Google Maps question. [13:13:50] Vera de Kok? [13:16:44] yes MaxSem [13:18:09] ryuch_: I think you are. :-) [13:29:07] nevermind, I'll write onto her talk page [13:30:34] ls [13:30:37] gah [13:30:38] sorry [13:49:41] @notify scfe_de [13:49:41] I'll let you know when I see scfe_de around here [13:49:48] @seen scfe_de [13:49:48] YuviPanda: I have never seen scfe_de [13:49:53] I *knew* it [13:49:58] I am spelling his nick wrong [13:50:10] anyone know what Tim's actual nick is? [13:53:04] @seen scfc_de [13:53:04] valhallasw: Last time I saw scfc_de they were quitting the network with reason: Quit: ERC Version 5.3 (IRC client for Emacs) N/A at 8/24/2013 1:57:20 AM (3d11h55m44s ago) [13:53:13] ^ I think that one [13:53:34] hi valhallasw, how are you today? [13:53:59] hi sumanah! Great - preparing a python course for students next week. [13:54:05] Cool! [13:54:12] I'd love to hear more [13:54:23] Are these experienced programmers who need to learn Python? [13:54:49] Some are, some aren't. It's a physics/astronomy group, and it's geared toward data analysis with python [13:55:43] which is an important step in the reproducability of science - if the analysis is well-defined, as it is in code, it's much easier to check results. [13:56:10] Even though the code would typically score F's for people in computer science ;-) [13:56:30] valhallasw: hah, I've been confusing e and c [13:56:30] ok [13:56:35] @seen scfc_de [13:56:35] YuviPanda: Last time I saw scfc_de they were quitting the network with reason: Quit: ERC Version 5.3 (IRC client for Emacs) N/A at 8/24/2013 1:57:20 AM (3d11h59m15s ago) [13:56:37] err [13:56:41] @notify scfc_de [13:56:41] I'll let you know when I see scfc_de around here [13:57:04] valhallasw: are you using ipython? [13:57:23] valhallasw: I'm a big fan of http://software-carpentry.org/ for teaching those folks [13:57:27] YuviPanda: ipython console for astronomy students, spyder for physics students [13:57:32] nice! [13:57:45] valhallasw: give them a peak of ipython notebook too :) [13:57:47] YuviPanda: but for a second course we will probably be switching to IPy 1.0+ notebook [13:57:52] hah! :D [13:57:56] nice, nice [13:58:05] I should setup a nice set of ipython notebook + publishing on toollabs [13:58:33] Not sure how to do that safely, though. [13:59:05] (I just upgraded to IPy 1.0 from 0.13. raw_input in notebooks ftw) [13:59:24] (I should try pywikibot in a notebook... O-)) [14:00:07] valhallasw: I've a proper system designed out. [14:00:37] valhallasw: start it as a SGE job, and then tunnel to it from your localhost :D [14:00:42] ah [14:00:49] valhallasw: and then with a bunch of ipython magics, be able to publish them at a static location [14:01:15] valhallasw: also more magics for easy SQL access, [14:02:43] pywikibot.input doesn't work in ipnb for now :( [14:03:27] probably because of the fancy unicode output, which patches up sys.stdin [14:05:49] valhallasw: heh, yeah. [14:05:54] valhallasw: if you enjoy doing that kind of teaching, you should definitely become part of the Software Carpentry posse [14:05:54] i've never used pywikipediabot [14:06:04] valhallasw: but pywikipediabot + ipynb sounds like a wonderful combo [14:06:29] sumanah: looks great :-) I'll take a look [14:06:45] :) [14:07:02] sumanah: hi :) [14:07:05] how are you [14:07:19] *grin* source control for physicists. Good luck with that :-) [14:07:32] hi Amir1_ , I am okay. How are you? [14:07:40] good [16:23:27] I'm a bit confused [16:23:49] which project would I use for mw development? [16:24:17] Mediawiki-dev? [16:25:19] normally I have a local instance [16:25:38] but I'd like to make use of http://netrenderer.com/ [17:07:37] also: can I make local changes and push them to review form an instance? [17:11:59] hmm, after more reading looks like it's not the right place [17:48:39] Yeay storms. ~40% packet loss atm. [18:24:32] !ping [18:24:33] !pong [18:29:06] YuviPanda: docker has been added as a new hypervisor in openstack nova [18:29:19] Ryan_Lane: oh wow. running on the metal? [18:29:21] should be released in the havana release [18:29:22] that'll be nice! [18:29:31] rather than running it on a vm [18:29:35] yep [18:30:42] <^d> Ryan_Lane: btw, yttrium is humming along just fine. [18:30:52] <^d> My little replication trick didn't work, but that's easy to switch later. [18:30:56] <^d> *ytterbium [18:31:01] Ryan_Lane: scfc_de almost got the package done last satruday, but he's disappeared after [18:31:03] <^d> stupid elements with similar names. [18:31:46] heh [18:31:48] <^d> The very first line of [[w:Yttrium]]. [18:31:52] YuviPanda: cool [18:31:54] <^d> "Not to be confused with [[ytterbium]]." [18:31:58] Ryan_Lane: do you know when andrewbogott_afk will be back? I could perhaps persuade him to pick up the packaging [18:32:03] <^d> Who the hell named these two things? [18:32:05] not sure [18:32:21] YuviPanda: it would be better to get andrew to do the wikitech interface [18:32:27] Ryan_Lane: that's true, yes. [18:33:14] Ryan_Lane: did we decide on having an API deamon vs having wikitech talk directly to redis? [18:33:20] ^d: yttrium \o/ [18:33:30] Ryan_Lane: also, should redis be the 'single point of truth'? i was wondering if we should store the data in mysql too [18:33:42] ^d: especially combined with barium and copper and oxygen [18:34:36] <^d> I was never good at chemistry. [18:34:59] <^d> Ryan_Lane: So, all we have left to do is...DNS changes + swap replication conf. When you wanna schedule that? [18:35:09] next week? [18:35:15] ^d: and basically: two people who discovered different atoms at the same place in Sweden, roughty 90 years apart [18:35:17] I guess we could do it this week [18:35:36] <^d> Just not tomorrow. [18:35:38] <^d> Thursday? [18:36:33] <^d> We should lower the TTL between now and then though. It's 1H now. [18:51:21] Ryan_Lane: Wtf is your avatar suppose to be? A squid hugging a shark? [18:52:02] * Damianz resists the urge to point out it sorta looks like a penis with legs [18:55:41] Damianz: you're not very good at resisting, are you? :P [18:56:07] What can I say, I'm a cryptic evil bastard [18:57:18] Damianz: hahahaha [19:09:03] <^d> Ryan_Lane: I'm going to send an e-mail to labs-l saying people can bug us about being renamed finally. [19:09:13] ok. sounds good [19:09:59] !email ^d ZOMG! I was renamed FooFooBunny! [19:11:49] <^d> Ryan_Lane: I'm not a bcrat, so I can't do renameuser on wikitech. [19:15:02] ah [19:48:48] welcome back, andrewbogott :) [19:49:01] Thanks! I'm very sleepy and disoriented :) [19:49:41] andrewbogott: hah! considered sleeping? :) [19:50:10] Tried, but… hoping to get at least partially in agreement with the sun. [19:50:13] Are you over your cold? [19:51:19] andrewbogott: mostly. My ears are okay, which is a relief. Just coughing now and then [19:51:41] andrewbogott: heh, my sleep cycle is currently 5PM to 6AM, so... Sun be damned :) [19:52:33] It is /way/ hotter in Minnesota than it was in HK. [19:54:19] andrewbogott: hah! [19:54:29] andrewbogott: wait, really? I thought it was supposed to be sno wfilled [19:54:55] 33C right now. [19:55:12] Won't have snow for a couple of months at least :) [19:55:25] andrewbogott: oh wow [19:55:39] was hotter yesterday, maybe things will mellow out a bit. [19:56:03] it was 34 earlier today. now at midnight, it's 28 :) [19:56:06] with 84% humidity [19:56:27] Yeah, so, about the same. I hope you have AC. [19:56:43] nope. [19:56:54] at least only 2 hours of power cut in the last week totally, so that's an improvement [19:57:15] wait, you don't have AC even when the power is on? That seems… difficult. [19:57:20] Esp. with a hot laptop. [19:57:40] hi andrewbogott :) [19:57:45] andrewbogott: ceiling fans. [19:57:47] * andrewbogott waves [19:58:13] andrewbogott: although it is still bad with the asbestos roof. I think I've just sortof gotten used to it [19:58:40] andrewbogott: am I allowed to pester you with 'work' stuff today? :) [19:59:23] Yep, I am supposedly working today, although may fall unexpectedly asleep at any moment. [20:02:09] silke_: OpenLayers or Leaflet as alternatives to Google Maps. LMAO Any idea of the level of copyfraud with map data? Its at least 40%. [20:02:43] LMAO? IDU! [20:02:59] what is LMAO? [20:03:09] Laugh My Ass Off [20:03:23] andrewbogott: ah, okay! so I was supposed to ping you about the wikitech interface to the dynamic http routing stuff (nginx+redis+lua) [20:03:58] Although the WMF is welcomed to hire me as some GIS guru [20:04:33] what is GIS? [20:04:37] :p [20:04:55] Geospatial Information Systems [20:05:25] andrewbogott: so wikitech will need to directly talk to redis on a configurable host, and let people add / modify domain / instance-name mapping as they wish, and only for things they've permission for [20:05:40] YuviPanda, is all the backend stuff in and working? [20:06:03] andrewbogott: pretty much, yes. [20:06:15] If you have a fully-formed vision of how this should work you can just rattle it off in an email and I can implement. Or we can discuss if it's unobvious... [20:06:17] andrewbogott: just not puppetized :P [20:06:29] Dispenser: What's the nature of that copyfraud? (I-m not a mappy person) [20:06:29] andrewbogott: as in, everything is puppetized except for the package [20:06:38] andrewbogott: wikistream-tools.wmflabs.org is running on it, for example [20:06:55] Ah, right, this is that nginx-extras thing? [20:07:04] Locating details from Google's Copyrighted satellite imagery [20:07:05] andrewbogott: only unobvious thing is wether redis should be the only place the mapping is stored, or if we should keep this in db too [20:07:12] Holiday had a pleasant amnesia-like effect [20:07:22] andrewbogott: yeah, that one. scfc_de started working on it and went far, but then he's disappeared [20:10:01] It possibly legal in the US and illegal in the EU. Locations are considered facts. [20:10:15] YuviPanda, I know close to nothing about redis. It's just a key/value store, right? [20:10:28] andrewbogott: pretty much, yeah. [20:10:37] andrewbogott: except that the 'values' can be anything - sets, hashtables, lists [20:10:45] andrewbogott: and you can perform atomic operations on the values themselves. [20:10:53] Is the data shared among multiple systems? [20:11:06] andrewbogott: in our setup, nope. it lives in a single machine [20:11:30] Dispenser: I'm not the expert here. But if I know there is one more house behind the forest and I go to map it, is it forbidden if Google told me the house is there? [20:11:50] andrewbogott: and while it isn't 'lossy' (unlike memcached - redis does persistance) - if it runs out of memory, it'll start chucking things out [20:12:13] silke_: My understanding is that the knowledge itself cannot be owned by google, only the representation. [20:12:31] So, for instance, you can 'trace' a map and your new map is maybe not a copyvio. [20:12:42] andrewbogott: so I am guessing, when I add a domain mapping in wikitech, we should 1. store it in the database, 2. then put that info on redis [20:12:44] But, I should let dispenser answer since I'm sort of guessing. [20:12:57] andrewbogott: once the db/redis thing is settled, I can write up a page on what the formats, etc are. [20:13:41] YuviPanda, it would be weird to have redis be the canonical representation of something wouldn't it? It's usually just a caching layer? [20:13:57] andrewbogott: yup, agreed. [20:13:58] in hindsight, I don't even know why I considered that :| [20:14:16] andrewbogott: let me write up a page? [20:14:26] Well, if you said, "no, it's totally normal to use redis for this kind of thing" then I would probably believe you :) [20:14:49] andrewbogott: some people do, indeed, use it like that. however, almost all of them will eventually be bit by it and then write blog posts saying how awful redis is ;) [20:15:00] silke_: US copyright ruling say facts are uncopyrightable, thus books or databases are uncopyrightable. However, at least some EU member countries (UK) allow databases to be copyrighted [20:15:05] OK, so, good argument for db then. [20:15:51] YuviPanda, should we wait for scfc_de to resurface or shall I start over on the packaging thing? [20:16:03] hm. [20:16:10] which db, though? [20:16:14] surely not wikitech's [20:16:16] Dispenser: What about your tools? You told me they can't move. I admit I can't remember what the do exactly. (I had a similar vacation amnesia to andrewbogott s) [20:16:21] andrewbogott: let's do the wikitech stuff first, and if scfc_de isn't back by then... [20:16:28] I don't see a problem with it being in redis [20:16:30] Ryan_Lane: why not wikitech? this is going to be part of a mediawiki extension... [20:16:46] *wave* Ryan_Lane :) [20:16:51] I'd prefer this be in an openstack service [20:16:51] Ryan_Lane: it can be trivially ddos'd by someone adding a lot, *lot* of entries. when MaxMemory is hit Redis will just throw out older keys [20:17:11] DDOS is the wrong word [20:17:15] YuviPanda: yeah, that makes sense [20:17:34] andrewbogott: it may make sense for this to be in the LBaaS part of nuetron [20:17:56] neutron* [20:18:28] I'm not sure if it fits well into the API, though [20:18:46] https://wiki.openstack.org/wiki/Quantum/LBaaS/API_1.0 [20:19:19] but the DB should definitely be separate from wikitech [20:19:41] YuviPanda: we can have quotas [20:19:54] Ryan_Lane: why *not* wikitech's db? [20:20:03] Man, every time I try to read about that it is so featuriffic that I nod off before absorbing anything. I'll try again :) [20:20:13] Ryan_Lane: more code :P plus if it is in the db it's automatically backed up, etc [20:20:24] redis is snapshotted [20:20:29] does it have no way to do backups? [20:20:33] silke_: I haven't started on the transition. I've also been around long enough to know WMF ultimatums aren't worth the bytes their stored in. Remember the Image filter referendum? [20:20:46] Ryan_Lane: you just backup the AOF file, and that's good enough. [20:20:52] YuviPanda: because I don't want the service linked to wikitech [20:20:54] Ryan_Lane: we shouldn't be using snapshots, should use AOF instead. [20:21:11] (the puppet changes currently use AOF, so it's all good) [20:21:17] eventually we want to have services available via CLI [20:21:25] I guess MW + oauth would allow that... [20:21:32] but keystone + oauth would too ;) [20:21:46] oauth is landing in keystone in the next release [20:21:46] Dispenser: I don't. But note that this is not a WMF ultimatum. It's mine - and those are reliable. [20:22:11] Ryan_Lane: me and Coren and scfc_de came up with a very nice solution that'll work for toollabs anyway :) We'll replace tools-webproxy with this at some point [20:22:28] YuviPanda: assuming we have quotas, it would be relatively difficult to overwhelm the memory [20:22:34] Lies! Lies all lies! [20:22:40] and we should also disable the feature that eject stuff from memory [20:22:43] What are we talking about? [20:23:05] Ryan_Lane: what else do you expect it to do when it runs out of memory? :P [20:23:08] die [20:23:18] Ryan_Lane: ah, I wonder if that can be... arranged [20:23:28] the amount of stuff we're going to actually keep in memory is incredibly small [20:23:40] if we run out of memory there's a serious problem [20:23:46] * Coren struggles with mysql and character sets. [20:23:59] and it sucks to need to talk to the db and to redis [20:24:49] Ryan_Lane: hmm, so that'll mean 1. quotas! 2. figuring out a way to make redis die when it has reached max memory 3. backups of redis' AOF / snapshots [20:24:54] vs 1. just add it to the db [20:25:02] #1 is likely enough on its own ;) [20:25:06] Auauauaugh! [20:25:12] quotas would be handled by wikitech [20:25:18] silke_: You're already at a disadvantageous with a suggesting system that isn't a marked improvement [20:25:20] where will it store the quotas? :P [20:25:25] until we had a proper API wrapped around it [20:25:26] in redis [20:25:26] * Coren beats mysql with a Unicode 1.0 reference manual. [20:25:34] and config [20:25:47] Ryan_Lane: or, I could write a small deamon that sits in front of redis? [20:26:10] well, then you'd need an api ;) [20:26:36] Dispenser: One thing you need to keep in mind, even if "isn't a marked improvement" was objectively true in the current situation, is that "runs at all" is a marked improvement over "turned off in 2014". [20:26:56] Ryan_Lane: for the edeamon to expose an API? that's easy enough :) [20:26:58] hah. this damn LBaaS api doesn't even define quotas [20:27:17] andrewbogott: let's ignore that stupid LBaaS spec [20:27:20] Ryan_Lane: I'm yet to see an API from OpenStack that didn't make me want to tear my hair out. Or maybe it is just the documentation... [20:27:39] Ryan_Lane: anyway, if I do the deamon, it can write to a local mysql/sqlite db [20:27:41] andrewbogott: we can wait till it lands, then write a driver if we want to use that [20:27:51] Coren: There may be other chapter interest in buy the Toolserver, perhaps the WMF too [20:28:39] Dispenser: The toolserver will lose database replication forever. Given the horribly old hardware, the by now broken stability, the non-open-source os and software running on it, that's simply not going to happen. [20:28:53] YuviPanda: I'd be happy with an API, but I'd very much want it to be OpenStack-like and I'd like for it to be able to support keystone auth in the future [20:29:11] Ryan_Lane, yep, I agree. [20:29:13] Ryan_Lane: I'm thinking simple REST-y stuff from flask. No Auth right now (just firewall to protect)? [20:29:14] I find the nova API to be relatively straightforward [20:29:34] YuviPanda: at least some simple form of auth would be nice [20:29:35] Dispenser: Tell us about your tools! Why do you think they don't run here? [20:29:48] YuviPanda: but really, this is just making things more complex [20:29:49] Because you pissed off Dab. [20:30:04] Ryan_Lane: keeping redis as a single source of data makes me feel very uneasy. [20:30:06] we can have wikitech write to redis directly for now and compute quotas itself [20:30:16] * Ryan_Lane nods [20:30:30] Ryan_Lane: plenty of 'redis sucks, it ate my data!' are born off these. [20:30:37] :D [20:30:51] well, if you're up to writing an API, that's fine with me [20:31:03] Then you told us that you were going to flush everything down the toilet for some idiotic open source philosophy [20:31:05] for auth, have it accept a token like an openstack service does [20:31:06] let me get to it then :) [20:31:26] Ryan_Lane: should be simple enough. Let me write up the API, and then you can tell me what's missing? [20:31:34] sounds good [20:31:48] * Ryan_Lane likes the idea of an API, rather than writing to redis directly [20:31:52] Then tell us we should be happy having to rewrite shit to make it work on your screwed up databases [20:31:58] YuviPanda: please version the API [20:32:04] Dispenser: If you find "open source philosophy" to be idiotic, then perhaps you aren't in the right place. [20:32:38] Dispenser: If you have actual technical issues with the databases that make them "screwed up", then please be explicit so they can be fixed. [20:32:51] Ryan_Lane: a simple /vX/ prefix. [20:32:51] it should be in the url, like: /v1/tools/service [20:32:56] hehe :P [20:32:57] yep [20:33:05] From what I understand User tables will not work [20:33:19] Dispenser: You understand incorrectly. They work very well tyvm. [20:33:20] YuviPanda: it may be good to see what's available in openstack-oslo [20:33:28] I'm here to write an open source encyclopedia. Not even RMS said everything must be open source. [20:33:29] [bz] (8NEW - created by: 2Chris McMahon, priority: 4High - 6enhancement) [Bug 53061] support Flow on beta cluster - https://bugzilla.wikimedia.org/show_bug.cgi?id=53061 [20:33:29] there's a decent amount of pre-written stuff there [20:33:50] (oslo is openstack common) [20:34:06] Dispenser: Perhaps if you took the time to figure out what tool labs /is/ before declaring how inadequate it is, you'd have a better spot to argue from. Right now, you're building a straw man. [20:35:10] Dispenser: do you have anything specific that needs to not be open source? [20:36:06] Ryan_Lane: looking at oslo now [20:36:15] if so, I'm sure people would be willing to help you switch to an open source alternative [20:36:43] * silke_ nods [20:37:06] and depending on what it is, there's a possibility of an exception of the open source only policy (though it's unlikely) [20:37:19] Ryan_Lane: I don't see anything that oslo provides that's going to be useful to me, though. sigh [20:37:26] but will keep an eye on it [20:37:28] YuviPanda: oh well. :) [20:37:43] Ryan_Lane: if it had something that does something Auth-y, I'd have used it :P [20:37:50] * Ryan_Lane nods [20:38:07] for the time being an "admin" token is totally acceptable for auth [20:38:13] since only a single service is going to talk to it [20:38:28] Ryan_Lane: right. [20:38:45] Ryan_Lane: wikitext also sucks for writing up API specs. [20:38:48] maybe project-level tokens would be nice [20:39:07] we could replace that with user-scoped tokens afterwards [20:39:14] My experience tells me these people to help me aren't going to write code [20:39:45] with project-level tokens the tools project would be able to modify its configs automatically [20:39:51] [bz] (8NEW - created by: 2Tilman Bayer, priority: 4Unprioritized - 6normal) [Bug 50556] Total number of Wikipedia articles is ca. 11 million too high - https://bugzilla.wikimedia.org/show_bug.cgi?id=50556 [20:40:06] [bz] (8RESOLVED - created by: 2Tim Landscheidt, priority: 4Normal - 6normal) [Bug 48626] Provide wiki metadata in the databases similar to toolserver.wiki - https://bugzilla.wikimedia.org/show_bug.cgi?id=48626 [20:41:15] Dispenser: probably a sane position to have [20:41:25] Dispenser: anyway, is there anything proprietary you are using? [20:41:56] Dispenser: Except that WMDE has specifically earmarked budget and staff (one dev, IIRC) exactly to help write code for transition when needed. (Correct me if I misunderstood, Silke) [20:42:19] Dispenser: If you want help you can talk to my colleague Johannes can help you. [20:42:33] There might be, I wasn't always too strict on writing my own code. [20:42:49] You'd have to contact him rather soon because I don't know what he is going to work on next year. [20:43:37] Dispenser: If you don't *want* help, I can't help though. ;) [20:45:04] What would make me happy is if we had a real time graph database [20:45:29] we have catgraph... [20:46:19] How close is it to real time? Minutes or days? Can it be made for pagelinks? [20:46:40] I have no clue. ask daniel kinzler about it [20:46:42] johannes is the expert [20:47:02] https://wikitech.wikimedia.org/wiki/Nova_Resource:Catgraph [20:47:07] he did it - all I know is that it keeps data in RAM to be fast [20:47:40] Dispenser: cat graph is running in tool labs already [20:48:18] People complain if changes don't show up in a couple minutes, it also reduces the usefulness. [20:48:21] äh... in Labs, right [20:57:41] Dispenser: Catgraph's last update is VALUE: 2013-08-27T20:53:01, some 4 minutes ago. I don't know what the interval is, however. [20:59:24] Better than the 2-5 days for http://toolserver.org/~lvova/ [21:01:01] the lvova tool will probably migrate soon. [21:01:18] They are in touch with Johannes [21:06:32] Ryan_Lane: sorry, my secondary internet went down (primary was already down, grrr) [21:06:54] Ryan_Lane: https://wikitech.wikimedia.org/wiki/User:Yuvipanda/Dynamic_http_routing/API [21:07:05] need to puth the token stuff in, will probably just add a cookie [21:08:14] YuviPanda: not a cookie :) [21:08:35] x-auth-token header, I think? [21:08:53] Ryan_Lane: I need to check what keystone does, and then mimic it [21:09:22] Ryan_Lane: but other than that? [21:09:31] andrewbogott: see link ^ [21:09:56] yeah, looking [21:10:05] I really need to go get lunch :) [21:10:07] back in a bit [21:12:30] pfft, lunch [21:23:23] [bz] (8ASSIGNED - created by: 2Tim Landscheidt, priority: 4Unprioritized - 6normal) [Bug 52559] Federate database centralauth - https://bugzilla.wikimedia.org/show_bug.cgi?id=52559 [21:36:22] Coren: http://tools.wmflabs.org/gerrit-reviewer-bot/ -- nothing in my php_error.log; index.php is owned by gerrit-reviewer-bot and there is no .htaccess [21:36:27] Coren: any idea what could be wrong? [21:37:04] valhallasw: I dunno, but it breaks things really badly. :-) Lemme go see what's up. [21:38:26] ooooh a meta table! awesome! :D [21:40:26]