[07:56:54] good morning [08:22:12] GerardM-: hello [09:17:09] Hoi Pyb [09:17:15] was in a meeting [09:17:42] http://ultimategerardm.blogspot.nl/2014/01/wikidata-discrimination-between-genders.html [09:20:26] what does wikidata's notion of gender look like? [09:22:59] a very long talk page ;) https://www.wikidata.org/wiki/Property:P21 [09:27:02] (03CR) 10Daniel Kinzler: [C: 04-1] "code readability issues" (036 comments) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106897 (owner: 10Addshore) [09:27:48] danbri Pyb the presentation of such info is used in the localisation of MediaWiki [09:28:14] in essence it is just a matter of using the same logci [09:28:39] you are asked in your preferences to indicate your gender [09:32:01] (03CR) 10Daniel Kinzler: [C: 032 V: 032] Fix JsonDumperGeneratorTest on Windows [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106733 (owner: 10Addshore) [09:32:29] (03Merged) 10jenkins-bot: Fix JsonDumperGeneratorTest on Windows [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106733 (owner: 10Addshore) [09:39:05] GerardM-: that's just for the UI though. I suppose danbri is asking about how the gender is represented in data items about people. [09:40:40] there is the property "sex (or gender)" P21 [09:41:11] https://bugzilla.wikimedia.org/show_bug.cgi?id=59989 [09:41:27] Afaik, there is currently no distinction between sex and gender (since that distinction does not translate well across languages and cultures) [09:41:34] yea, P21 [09:41:57] Qualifiers can/should be used in "non-standard" cases, e.g. https://www.wikidata.org/wiki/Q298423 [09:43:00] there we have P21 -> "male" (instance of sex), P21 -> "transgender female" (instance of gender identity), and P21 -> "female" (instance of gender identity) [09:43:09] that'S actually a pretty nice solution, i think [09:43:41] danbri: --^ [09:43:45] but it has no application [09:43:51] and that is what I care about [09:44:23] it doesn't? [09:44:26] no [09:44:38] where is it applied ? [09:44:56] you mean, used inside of wikidata? [09:45:02] yes [09:45:03] it isn't, since the software doesn't know or care about it [09:45:17] it could and it should [09:45:46] the software shouldn't know about any specific properties, no. not wikibase itself. [09:45:57] software built on top, sure. but not wikibase itself. [09:46:12] why ? [09:46:27] because hardcoding the meaning of properties is asking for trouble [09:46:29] and how do you make it stick if it is not in the base software [09:46:40] "make it stick"? [09:46:52] not hardcoding is asking for trouble as well [09:47:40] how do you apply gender if you cannot rely on what gender is ? [09:47:46] it means not taking shortcuts. that makes things less convenient, but more flexible and consistent. [09:48:13] the problem is that there is no concise definition of what gender is. hardcoding that in the software would be silly. [09:48:19] convenient for who ? flexible and consistent at what price ? [09:48:38] from a grammar point of view there is [09:49:21] the price of the software not knowing what "subclass" or "instance" means, for instance. which is quite a high price. but means that the software is not imposing any ontological framework. which is quite an important dfistinction from many systems, and allows for a lot of freedom [09:50:09] language and the vagaries of language are not part of an ontological framework and do not rely on it either [09:51:05] languages actually impose an ontological framework (genders, moods, word types, etc - an ontology) [09:51:06] when applied the rules of languages are the same for any ontology [09:51:23] the option is to apply it or not [09:51:35] but there is nothing more [09:52:04] my point is: the notion of "gender" is defined in the content. so any code using it would have to be defined as content - site level JS, gadgets, whatever. [09:52:29] hardcoding it in the core software would be a mistake, since that would remove freedom/flexibility from the content level. [09:52:46] it wopuld prescribe some (at least grammatical) notion of gender [09:53:15] (or of class/instance relationships, or geographic inclusion, or whatever other convenient properties you'd like the software to know about) [09:53:20] ... I asked earlier ... convenient for who ? flexible and consistent at what price ? [09:53:32] i'm not saying it's bad to have software know or use this. i'm saying it would be bad if the wikibase core software would do that [09:53:54] it is something that is either on or off [09:54:04] that is all [09:54:26] well, it would be convenient to be able to search and navigate based on subclass relationships, for instance. the price would be imposing an ontological framework, instead of leraving that to the community, to be defiend on the content level. [09:54:54] DanielK_WMDE it is NOT about subclass et all [09:54:59] it is about language [09:55:24] the whole notion of classes is not relevant [09:56:12] it's just another example of the same principle: knowledge about the "meaning" of properties does not belong into the core software. Any such knowledge needs to be defiend and maintained on the content level. Doing otherwise would be a "category mistake". [09:56:59] but you do not address the point that language is the same NEVER MIND the structure of an ontoloty [09:57:08] i'm just using the "class/instance" thing as another exampel oif a very convenient property people might want the software to know about. far more powerful and convenient than gender. and still we decided not to support that on the level of the core software. to retain complete freedom on the content level. [09:57:52] ... it does not apply because language is not defined by classes and other ontological constructs [09:57:58] GerardM-: it'S not about the structure of the ontology. it's about keeping any knowledge about "meaning" out of the software, as a fundamental design choice. [09:58:39] all properties are the same to the sfotware. it does not and should not know any thing abotu any specific property, beyond what is explicitly defined in the defintion of that property. [09:58:57] ok ... leave this argument for a moment ... how would you apply gender when it is NOT part of the base functionality [09:59:05] if you want gendered labels based on the content, you'Äll have to build software on top of wikibase, not build this into wikibase. [09:59:24] site js would be the obvious choice. [09:59:27] that does not answer the question [10:00:12] would that affect the API ? [10:00:31] no [10:00:37] than it is no good [10:00:48] o_O [10:01:01] you want consistent results [10:01:07] indeed. [10:01:20] and when consistent results are not provided it is not good enough [10:01:46] GerardM-: the results are consistent: the result from your example is Q33999 [10:02:35] and the label returned is "actor / actress" [10:02:56] which is no good for Mrs Zeta-Jones as she is clearly an actress [10:04:01] it should be actress because actress has a statement for "female form" or whatever label is used [10:04:21] sex female would be better [10:04:33] so how would the API return it ? [10:04:46] would you have that logic also at the other end ??? [10:04:50] GerardM-: actually, that is semantically wrong. the classes (or sets) "actor" and "actress" are not identical, nor are they the same as "actor/actress" [10:05:07] wikidata (on the content level) decided not to make the distinction. that's it. [10:05:26] the current label for q33999 is "actor /actress" [10:05:34] yes. [10:05:47] and having this further subdevided is not needed - serves no purpose [10:05:53] because it does not translate [10:06:12] so, making the disctinction in the API would be simply wrong. [10:06:17] it would lead to inconsistent results [10:06:38] the data structure doesn't make the distinction, so the API doesn't. [10:06:40] returning a label that is wrong is also stupid [10:06:52] so the data structure needs to change [10:06:56] cheers for merging https://gerrit.wikimedia.org/r/#/c/106733/ DanielK_WMDE :D [10:06:57] it'S nto wrong. it's the accurate label: the class/set of all actors AND actresses. [10:07:15] but it is data that is not usable [10:07:16] it's not the class of actors, not the class of actresses. it'S the union of the two. [10:07:46] in nomen clature there are the splitters and the mergers [10:07:57] I hate splitting for no obvious reason [10:07:58] sure it'S usable. just not as conveniently and nicely as you would like. which is the price of having a generic system instead ofa specific one. [10:08:11] linguistically you are insisting on splitting [10:08:24] i am? how? [10:08:57] because you want to have every profession that has a different gender have its own item [10:08:59] STUPID [10:09:08] so how would you deal with plural ?4 [10:09:44] as to gender ... some language have other requirements ... how do you deal with that ? [10:09:52] no, i don't. [10:09:53] or no requirements ? [10:09:53] GerardM-: you want a nice way to show the "correct" label to the user. But that "correct" label doesn't reflect what's actually defined in the data structure. it applies extra implicite knowledge (a female member of class q33999 is called an "actress", a male member an "actor") [10:10:23] that knowledgfe is *content*, and should be treated and maintained as *content*. [10:10:24] My point is that the data structure should allow for it [10:10:33] the software can't and shouldn't know about that [10:10:50] that is fine as long as the result provided is correct [10:10:57] then the data structure would have to know about some notion of gender [10:11:13] the result procided *is* correct. it's just not convenient for your use case. [10:11:24] linguistically it is not [10:11:44] who do we do it for ? [10:11:55] and what is ultimately what Wikidata stores ... [10:11:56] but wikidata isn't a translation system, not a dictionary. it doesn't deal in words but in concepts. [10:12:02] it is the Q and P numbers [10:12:08] wikibase support for linguistic entities is a couple of years off, sorry [10:12:36] it does not need to be [10:13:21] ... technically there are several reasons why we want statements for labels [10:13:26] ok, here's a proposal: we want to support claims on properties (i.e. "this property is deprecated" or "this property only applies to instances of cars") [10:13:44] ok [10:14:04] the first is a good point [10:14:05] once that works, you can have "the label for femal members of this class is" as a multi-lang text (once we have multi-lang text) [10:14:08] the second is not [10:14:40] what, the "cars" example? just a random example. people can come up with whatever they want to say about properties. [10:15:08] specifying the range and domain of properties is pretty standard, so i expect that to happen. [10:15:09] the deprecated example is not data related [10:15:41] actor and actress are both labels for q33999 [10:16:05] they are NOT multi lingual texts [10:16:21] no idea what you mean by that, but that's not the point: the point is once you can make claims about statements, you can assign extra info, e.g. gender specific labels. Thislets software that understands these properties use gender specific labels. [10:16:34] but wikibase wouldn't know about them. it's application level knowledge. [10:17:10] Daniel ... actor is not a statement ... it is a label [10:17:25] hm, actually... i got off track here. you don't need claims about properties to do what i was thinking of. [10:18:10] GerardM-: i know that, i wrote the code :) [10:19:06] GerardM-: but the currect label for the class is not a gendered one, because the class isn't genedered. Either labels would have to be complex linguistic objects instead of strings. or you use statements to express what you are looking for. [10:19:47] imaging q33999 having a property called "term for femal members", with a multi-lang value which would be "actress" in en, "Schauspielerin" in de, etc [10:20:10] that would provide the info you want to any application that knows about the special "term for femal members" property [10:20:19] by wikibase sure shouldn't have that knowledge [10:20:35] so how would you then do it for plural ? [10:20:52] it really makes sense to have statements on labels [10:20:54] i would wait for actual support for linguistic objects in wikibase [10:21:07] but we need it already [10:21:25] this use case (using linguistic objects as labels for data items) was actually considered in the design (even if the proposal doesn't talk about it much) [10:21:34] then fake it. [10:21:46] either wait for *proper* *generic* support, or fake it [10:21:55] don't compromize the system. [10:24:25] GerardM-: can we agree that it would be quite cool to have linguistic objects as labels, so we have machine readable information about all the forms of the lables, plus pronunciations, etc? Can we also agree that this would be quite a compelx piece of data structure, if we want to get it right? [10:24:37] that's exactly what the "wikidata for wiktionary" proposal covers. [10:24:41] proper label support is how you apply deprecated as well [10:25:04] though the details of how items get associated with expressions are still a bit vague [10:25:26] that is the lemon in lemon [10:25:31] hm... i disagree about deprecating per label. but whatever, i need to get back to code review :) [10:25:52] it does not consider how labels and linguistic content are connected [10:25:57] indeed. the model is very simmilar to Lemon [10:26:07] and to OmegaWiki [10:26:11] i know [10:26:15] not a conisidence [10:26:23] * DanielK_WMDE can't type today [10:26:29] * DanielK_WMDE can't type, period [10:26:47] :) DanielK_WMDE thanks :) [10:27:41] GerardM-: go ahead and blog about how the "wiktionary/wikidata" thingies could be used to fix the gender issue (and the genitive issue, and the plural issue, etc) [10:27:59] GerardM-: cheers :) [10:29:43] * hoo waves [10:29:56] hey hoo [10:32:25] (03PS4) 10Addshore: Merge References for Statements in wbmergeitems [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106897 [10:34:20] (03CR) 10Daniel Kinzler: [C: 04-1] "Looks to me like a stripped down version of the tests got committed accidentally." (034 comments) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/96268 (owner: 10Addshore) [10:36:54] (03CR) 10Daniel Kinzler: [C: 032 V: 032] Don't get extra languages for anon users [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106689 (owner: 10Tobias Gritschacher) [10:37:15] (03Merged) 10jenkins-bot: Don't get extra languages for anon users [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106689 (owner: 10Tobias Gritschacher) [10:39:20] (03CR) 10Addshore: Merge References for Statements in wbmergeitems (035 comments) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106897 (owner: 10Addshore) [10:39:57] DanielK_WMDE: https://gerrit.wikimedia.org/r/106905 Opinions? Especially about the class rename/ namespace change [10:43:21] (03PS3) 10Daniel Kinzler: (bug 47070) No import of wikitext to entity NS [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100565 [10:48:14] (03PS15) 10Tobias Gritschacher: Use snake_case for methods and variables [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 (owner: 10Aaron1011) [10:49:56] (03CR) 10Daniel Kinzler: [C: 04-1] "minor questions re naming, documentation" (033 comments) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106905 (owner: 10Hoo man) [10:51:10] (03PS5) 10Addshore: Merge References for Statements in wbmergeitems [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106897 [10:52:15] hoo: looks like an improvement, but i have some questions (see comments9 [10:53:27] DanielK_WMDE: ok, will reply... it's basicly a split out commit... there are going to be two more... while it works, I want the full stack to go live at one point cause only at that point we will have integration tests [10:54:05] maybe it's best to review the full stack once it's up, then [10:54:54] might be... although that change is harmless alone as it only changes the style and implements things not used by default (at that point) [10:55:32] yea, but still worth thinking about whether it's moving in the right direction [10:58:33] DanielK_WMDE: I think it makes sense seen in the whole context... in the end we will have two Lua Libraries mw.wikibase and mw.wikibase.entity (with two implementation classes). And the legacy format will stay in place while we will have new mw.wikibase function which will return a non-legacy mw.wikibase.entity [10:58:45] + a [10:59:23] yea, think i see. sounds good. [11:02:40] (03PS14) 10Tobias Gritschacher: Migrate simple RSpec browser tests to Cucumber. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100591 (owner: 10Vldandrew) [11:10:35] (03PS13) 10Addshore: Add ungrouped list api parameter [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/96268 [11:10:43] (03PS14) 10Addshore: Add ungrouped list api parameter [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/96268 [11:17:41] (03CR) 10Hoo man: Refactor the PHP part of the Lua library (033 comments) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106905 (owner: 10Hoo man) [11:18:35] Tobi_WMDE: Didn't you set up a bot which should run unit tests for every commit like WMF jenkins used to? :P [11:21:14] hoo: nope. what I did last week was setting up a user on gerrit which has the necessary rights to receive streaming events from gerrit. [11:21:51] with that user we would be able to run jobs on every commit [11:22:13] ah :) [11:22:28] " ! [remote rejected] HEAD -> refs/publish/master/Lua (missing Change-Id in commit message footer)" wtf, git [11:22:33] hoo: writing this job (setting up mediawiki, databases, etc.. installing extension, running tests..) still to do [11:23:25] hoo: if you want to give that a try, just tell, I can give you access to our jenkins on cloudbees [11:23:43] DanielK_WMDE: Any idea why git tells me " ! [remote rejected] HEAD -> refs/publish/master/Lua (missing Change-Id in commit message footer)" although head is on the commit already in gerrit? [11:25:04] as far as I know katie and hashar spoke about an own jenkins instance for wikidata on labs last week, I'm not sure about the status of that right now.. [11:25:33] Yeah, I think that will be part of the already existing Wikidata project on labs [11:25:56] I don't have access to that through... [11:26:13] I would like to have a setup very similar to Travis... maybe we could even use the same scripts [11:28:25] hoo: you think that's possible on jenkins? [11:28:58] Tobi_WMDE: Sure... why not ;) [11:29:37] hoo: I mean using the same scripts :) [11:29:37] you can use whatever arbitrary code you want in there... only downside is that things might get ahcky at some point :P [11:29:41] that would be nice [11:30:02] I guess so, I use derivative ones for quite a while on my own tests VM [11:30:13] * test VM (only one) [11:30:29] if you want to play around I can send you access to cloudbees.. the jenkins there is already hooked up with gerrit and can run jobs on every commit [11:30:49] (03PS7) 10Addshore: Die if wbeditentity tries to clear from old revid [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98795 [11:31:04] (03PS5) 10Hoo man: Refactor the PHP part of the Lua library [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106905 [11:31:06] (03CR) 10jenkins-bot: [V: 04-1] Refactor the PHP part of the Lua library [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106905 (owner: 10Hoo man) [11:31:14] (03CR) 10Addshore: [C: 04-1] "still needs tests" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98795 (owner: 10Addshore) [11:32:47] wtf... how can a change based on master not merge with the current repo [11:33:04] :D [11:33:11] havving gerrit / git / jenkins fun? :D [11:33:42] addshore: Yeah :/ These move to new * components changes seemed to have messed something up for me [11:33:59] I even had to cherry pick that change out of its original branch first... [11:34:09] (to only get it to upload on gerrit) [11:35:28] will investigate this later ... [11:36:04] DanielK_WMDE: https://gerrit.wikimedia.org/r/106905 if you have an idea why this (although it's rebased on master) doesn't merge with it, please tell me [11:36:06] away for now [11:50:35] catching up, thanks for the gender link GerardM- [11:52:01] so wikidata says that a human can only be one of [male (Q6581097), female (Q6581072), intersex (Q1097630), genderqueer (Q48270)] [11:52:33] if someone's gender is marked as genderqueer or intersex, does that help you decide whether to call them 'actor' or 'actress'? [11:53:07] FWIW here's what I wrote for the FOAF spec a while back, http://xmlns.com/foaf/0.1/#term_gender [12:02:06] the way MediaWiki works is that it asks you how you want to be addresed [12:02:57] Typically there are two formats for addressing a person male and female.. The language defines what is default [12:03:09] default is also for when the gender is not known [12:03:35] (03PS1) 10Zfilipin: Migrate simple RSpec browser tests to Cucumber. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107129 [12:05:45] (03Abandoned) 10Zfilipin: Migrate simple RSpec browser tests to Cucumber. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100591 (owner: 10Vldandrew) [12:06:46] (03CR) 10Zfilipin: "Moved from https://gerrit.wikimedia.org/r/#/c/100591/" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107129 (owner: 10Zfilipin) [12:07:33] (03PS8) 10Addshore: Die if wbeditentity tries to clear from old revid [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98795 [12:15:31] (03PS1) 10Addshore: Remove all use of getSimpleSiteLink method [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107130 [12:22:58] (03PS2) 10Daniel Kinzler: Support for legacy schema in TermSqlIndex [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106721 [12:23:00] (03CR) 10jenkins-bot: [V: 04-1] Support for legacy schema in TermSqlIndex [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106721 (owner: 10Daniel Kinzler) [12:24:02] (03PS6) 10Daniel Kinzler: (bug 56711) use prefixed IDs in wb_terms table. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101197 [12:24:29] (03PS3) 10Daniel Kinzler: Support for legacy schema in TermSqlIndex [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106721 [12:29:40] (03CR) 10Daniel Kinzler: Die if wbeditentity tries to clear from old revid (031 comment) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98795 (owner: 10Addshore) [12:30:57] (03PS1) 10Addshore: Remove all use of deprecated addSimpleSiteLink [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107132 [12:32:11] (03CR) 10Addshore: Die if wbeditentity tries to clear from old revid (031 comment) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98795 (owner: 10Addshore) [12:45:13] (03Draft1) 10Addshore: Remove ALL @file and @ingroup comments [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107135 [12:50:33] (03CR) 10Daniel Kinzler: Die if wbeditentity tries to clear from old revid (031 comment) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98795 (owner: 10Addshore) [12:50:36] (03Draft2) 10Addshore: Remove ALL remaining verbose license headers [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107139 [12:53:14] (03PS9) 10Addshore: Die if wbeditentity tries to clear from old revid [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98795 [12:53:26] (03CR) 10Addshore: Die if wbeditentity tries to clear from old revid (031 comment) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98795 (owner: 10Addshore) [13:13:46] (03Draft1) 10Addshore: Remove ALL unused imports [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107144 [13:30:46] (03Draft1) 10Addshore: Fix undefined classes [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107145 [13:30:49] JeroenDeDauw: meeeerrgeggeeee ;p ^^^^ [13:39:03] aude: DanielK_WMDE: Lydia_WMDE: who wants to join the daily at 15:00? [13:39:15] * aude waves [13:39:26] i doubt lydia will join unless she is very jetlagged [13:39:59] (03CR) 10Tobias Gritschacher: [C: 032 V: 032] Migrate simple RSpec browser tests to Cucumber. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107129 (owner: 10Zfilipin) [13:40:26] (03PS16) 10Tobias Gritschacher: Use snake_case for methods and variables [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 (owner: 10Aaron1011) [13:40:58] she will be in the US right ? [13:41:48] aude: got it! [13:43:16] (03PS6) 10Addshore: Refactor the PHP part of the Lua library [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106905 (owner: 10Hoo man) [13:43:53] (03CR) 10Addshore: "Rebased and jenkins seems to like it now!" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106905 (owner: 10Hoo man) [13:44:27] * aude reviewing addshore's stuff [13:44:39] :P [13:44:42] aude: any idea when we're able to update beta again? think the currently deployed version is from jan 8th [13:45:04] Tobi_WMDE: no idea [13:45:19] aude: what's the blocker there? [13:45:23] once we get the build running? :P [13:45:32] make sure it works with localisation cache [13:45:56] ah, ok. but we tried the single-repo there and it worked, right? [13:46:00] yes [13:46:13] localisation cache knows about submodules, fine [13:47:03] aude: ok, so we would make the Wikidata.git repo use submodules again and that would still work then? [13:47:16] it uses submodules now [13:47:19] but we turned submodules off for Wikidata.git [13:47:28] we enabled it again [13:47:29] we turned the auto updating off [13:47:30] oh, ok [13:47:33] oh :O [13:47:36] what addshore says [13:47:59] * aude shall first work on jenkins :) [13:49:08] if anyone cares, https://github.com/wmde/WikidataBuilder/pull/5 (doesn't fix localisation) [13:49:15] but changes i think needed [13:50:34] aude: don't we have a batch script in favor of the node thing? [13:50:39] * addshore goes to make github notify irc now [13:50:49] Tobi_WMDE: no idea [13:50:50] do we? [13:50:55] where? [13:50:55] addshore: good thing! [13:51:09] aude: JeroenDeDauw mentioned that IIRC [13:51:14] huh [13:51:15] JeroenDeDauw: ? [13:51:16] ^^ [13:51:38] ??? [13:51:53] or we can make one [13:51:58] it's just composer install [13:52:36] aude: yes, I'm wondering why we would need this node script [13:53:02] we really don't [13:53:48] aude: all we really need is a composer install and pushing that to the wikidata.git [13:53:55] and make localization work [13:53:57] we just need to make sure it uses the right options [13:53:59] without submodules [13:54:25] --prefer-source --optimize-autoloader [14:00:26] * aude waiting [14:09:29] JeroenDeDauw: i can has access to the github datavalue stuff again? [14:14:31] * aude getting segfault in phpunit [14:15:19] :O [14:15:44] it's good! maybe i can find the cause [14:15:59] (03CR) 10Tobias Gritschacher: [C: 032 V: 032] Use snake_case for methods and variables [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 (owner: 10Aaron1011) [14:16:00] it's at the end of the client tests [14:16:43] aude: I've seen segfaults in the phpunit tests on travis.. [14:16:49] yes [14:20:40] [02wmde/wikidata-analysis] 07addshore pushed 031 commit to 03testircnotifcation [+0/-0/±1] 13http://git.io/Cu863Q [14:20:41] [02wmde/wikidata-analysis] 07addshore 03b230d44 - Test commit to test irc notifications DNM Do not merge [14:20:44] woo :P [14:20:58] yay! [14:21:08] addshore: \o/ [14:23:29] hmm, not working for pull requests for some reason [14:32:17] aude: which PHP version are you running? 5.3.x [14:32:18] ? [14:32:30] (03CR) 10Aude: [C: 04-1] Fix undefined classes (031 comment) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107145 (owner: 10Addshore) [14:32:33] addshore: Thanks for rebasing, dunno what messed up over there [14:33:09] i don't know if it's the same issue [14:33:26] rather an issue with wikibase (maybe addshore's patch) and new data model [14:34:05] mh... but still a segfault is most likely not our issue, but a PHP bug (which we should isolate and report/fix) [14:34:14] Are you running 5.3? [14:34:17] 5.3 yes [14:34:31] it could be just the test exiting [14:34:34] the latest an greatest .28, I guess [14:34:47] Program exited with code 0377. [14:34:55] mh... shove it into gdb and have a look [14:35:06] PHP 5.3.27 (cli) (built: Jul 19 2013 18:25:06) [14:35:10] doing [14:35:19] that's how i found the issue above, but think there might be more [14:41:37] [13wikidata-analysis] 15addshore closed pull request #5: This is a test pull request (06master...06testircnotifcation) 02http://git.io/5CRB2A [14:41:47] woo! [14:42:20] github-wmde is so unsocialable [14:42:32] pick a new name? :P [14:42:42] it should stay here :) [14:42:50] ohhhhh :p [14:42:51] like AsimovBot [14:42:59] dont think I can make it do that :P [14:43:03] heh [14:44:00] Tobi_WMDE: could you give me access to the DataValues repos so I can do the same? [14:45:33] 01[13wikidata-analysis01] 15addshore 04deleted 06testircnotifcation at 14b230d44: 02http://git.io/kMIC9Q [14:47:12] also Tobi_WMDE decide which of the hooks we want enabled fffor the notifications from http://developer.github.com/v3/activity/events/types/#pullrequestevent [14:47:49] currently we have "push", "pull_request", "commit_comment","pull_request_review_comment" [14:48:17] all of those would be similar to what grrrit-wm does [14:48:30] addshore: What is it with you and IRC bots for Git [14:48:35] :p [14:48:35] if it's not too much noise [15:03:21] (03PS2) 10Addshore: Fix undefined classes [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/107145 [15:03:23] aude: ^^ =] [15:04:00] addshore: ok, although i think there are more issues [15:04:13] really? I wonder why that test didnt fail for me also! [15:04:17] shall run on travis also [15:04:19] client only? [15:04:28] ahhh!! [15:04:29] yus! [15:04:45] yus a run on travis would be good [= [15:05:27] the first 4 changes should all be fine :P [15:05:43] they seem fine but i'm runnign tests on all patches together [15:05:48] easier this way :) [15:06:41] if i point phpunit to my client directory, then it passes [15:06:49] --group Wikibase exits [15:09:38] hmm, I was using --group Wikibase and they passed, but I dont have a client setup [15:09:52] that would explain [15:09:58] indeed! [15:10:22] i am trying master [15:27:02] [travis-ci] filbertkm/Wikibase#116 (master - f3d17fa : Zeljko Filipin): The build is still failing. [15:27:02] [travis-ci] Change view : https://github.com/filbertkm/Wikibase/compare/7eacbeb8c9d1...f3d17fa587f4 [15:27:02] [travis-ci] Build details : http://travis-ci.org/filbertkm/Wikibase/builds/16872419 [15:27:43] gah [15:28:13] Undefined index: bot on php 5.4 [15:29:04] is anyone here responsible for running the wikidata production server? [15:29:41] jamesclarke: Mostly developers around here... better just ask your question [15:30:08] I have an issue where entities with lots of data (e.g., Q148) are returning server errors [15:30:12] when requesting json [15:30:34] https://www.wikidata.org/wiki/Special:EntityData/Q148.json [15:30:34] HTTP/1.1 503 Service Unavailable [15:31:25] whereas smaller entities work fine [15:31:26] 10:30:03 ~/projects$ curl -I https://www.wikidata.org/wiki/Special:EntityData/Q42.json [15:31:26] HTTP/1.1 200 OK [15:32:02] aude: ^ [15:32:11] hmmm [15:32:51] https://www.wikidata.org/w/api.php?action=wbgetentities&ids=q148&format=json works [15:32:52] (03CR) 10Daniel Kinzler: Merge References for Statements in wbmergeitems (031 comment) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/106897 (owner: 10Addshore) [15:33:46] that needs further investigating... jamesclarke do you want to open a bug? ;) [15:34:08] that's a software thing, not really related to the actual hosting of Wikidata [15:34:44] sure; where should I submit bug? [15:34:56] Also is api.php documented and available for public consumption? [15:35:11] "wikidata-repo" component in bugzilla [15:35:23] https://bugzilla.wikimedia.org/enter_bug.cgi?product=MediaWiki%20extensions [15:35:31] you can use api [15:35:39] https://www.wikidata.org/w/api.php has documentation [15:35:40] jamesclarke: Sure, it's both documented and you're free to use it! :) [15:36:10] maybe there is some timeout setting or something for entitydata [15:36:37] mh, it's basically using hte API formatters for the data, so I'm not sure why this is failing [15:36:52] api request was quick for me, so don't know [15:37:09] DanielK_WMDE might have an idea [15:37:49] [travis-ci] filbertkm/Wikibase#117 (undefinedclasses - ce42bc1 : addshore): The build passed. [15:37:49] [travis-ci] Change view : https://github.com/filbertkm/Wikibase/compare/01de625b1a9f^...ce42bc1aceeb [15:37:49] [travis-ci] Build details : http://travis-ci.org/filbertkm/Wikibase/builds/16872488 [15:38:08] mh... I think it saw some recent changes... I poked at it about a month ago to investigate some unit test stuff or whatever [15:38:11] jamesclarke: it's a timeout, yes, but an unspecific one. *something* is taking too long [15:38:29] if you file a bug report listing a few queries that trigger this, i'll investigater [15:39:00] ok will do [15:39:03]