[17:08:01] (03CR) 10Hoo man: [C: 031] ""view/resources/resources.php" looks rather pointless now, but I guess you plan to add more to that." [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200860 (owner: 10Adrian Lang) [17:09:20] (03CR) 10Hoo man: "recheck" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/198427 (https://phabricator.wikimedia.org/T70380) (owner: 10Daniel Kinzler) [17:17:37] bye hoo ;) [17:17:41] (03CR) 10Hoo man: [C: 04-1] Refactoring dispatchChanges.php (031 comment) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/198427 (https://phabricator.wikimedia.org/T70380) (owner: 10Daniel Kinzler) [17:18:37] cu benestar :) [17:29:29] hello [17:29:38] hi stoopkid :) [17:30:13] im interested in helping out for the design/dev of the wikidata UI [17:30:35] ping Lydia_WMDE [17:31:39] stoopkid: The UI is a place that's rather hard to start in... it might be better to look at some other area first. [17:32:07] hoo: but UI is the place where you need most help, Lydia said [17:32:27] there's so much to do in the UI [17:32:43] e.g. extend geo coords with the possibility to choose the planet, etc. [17:32:43] hoo: what's difficult about starting with the UI? [17:32:46] Well, it's certainly one of the places that lack a lot developer time, yes [17:33:02] and there are many tasks which are rather indepedent of other tasks [17:33:20] add calendar models to time [17:33:31] stoopkid: The UI is very complex and hard to understand... even I don't really get how everything works together there exactly. [17:33:34] heck, add a more button to the language box [17:34:01] But yes, there are some tasks that are potentially doable without having to get to know the whole thing first [17:34:07] Like the one ULS uses? https://www.mediawiki.org/wiki/Universal_Language_Selector/Design/Interlanguage_links [17:35:16] Nemo_bis: I was just thinking of a [more] link at the bottom of the "In other languages" box, which literally displays more languages than the configured few [17:35:29] and then an [even more] button to get all [17:35:43] (could also be called [all]) [17:35:52] there's a task for that in phabricator [17:36:01] dennyvrandecic: and how would you select the few languages? [17:36:17] all languages that have a label, description, or alias at all [17:36:19] I can understand "all" [17:37:08] Still, this is something that would benefit from sharing with ULS. [17:37:36] maybe, but it would also benefit from being done [17:38:55] Maybe it's easier by using ULS. Just saying :) [17:39:21] maybe :) up to whoever does it [17:39:54] stoopkid: anything in particular you want to work on? [17:41:30] dennyvrandecic: i have ideas for things, but it will take me some time to go over the code-base and get familiar with it [17:42:06] you can also check phabricator. it probably has tasks marked as "pick up for volunteers" and "easy to do" [17:44:07] dennyvrandecic: https://phabricator.wikimedia.org/tag/wikidata/?order=natural here? [17:45:13] Lydia_WMDE: how do you use phabricator to find tasks marked for volunteers and / or easy? [17:47:25] stoopkid: here's a possibility https://phabricator.wikimedia.org/tag/need-volunteer/ but I guess it gets better than that [17:57:03] dennyvrandecic, Stryn: https://phabricator.wikimedia.org/maniphest/query/Ob8tyN8BM.gq/#R but beware the list is not curated as much as we would like, so there might be tasks on it that nobody looked at for a year, ask here if in doubt about that [17:57:18] why me? :O [17:57:30] err i meant stoopkid [17:57:32] because you autocompleted wrongly for stoopkid [17:57:36] ;) [17:58:12] there is probably a setting to make irssi behave in a sane way [18:01:59] *sigh* there is not :( [18:03:18] SMalyshev: hey! I'm online for a couple of hours. I'm finishing tests for the RdfBuilder split, and will then follow up on comments/rebases/reviews. [18:03:32] DanielK_WMDE_: cool [18:03:47] Is there anything that keeps us from merging the move of RdfWriter to the "purtle" component? [18:04:50] matej_suchanek: sorry back now [18:04:56] anyone here in touch with the inventaire folks? [18:05:06] Eloquence: yes me [18:05:16] Lydia_WMDE, do you know if they're contributing back yet? or just querying for now? [18:05:26] querying only for now but more is planned [18:05:37] thanks [18:05:50] Lydia_WMDE: pinged you just because of the talk above [18:05:56] reading :) [18:06:51] how bout this: https://phabricator.wikimedia.org/T57549 [18:06:59] looking [18:07:01] "add a new datatype for geoshapes" [18:07:33] stoopkid: there are quite a few open question about the feasability [18:07:43] if you're ok with that it'd be awesome [18:07:48] but be aware that this is rather big [18:07:59] Lydia_WMDE: what are the questions about the feasibility? [18:07:59] it's nice however because it is very independent [18:08:16] mostly about item size as lined out in the comments in the ticket [18:08:32] ie can we store the json blobs or will they get too big [18:09:16] if you want to start with something smaller then i think what dennyvrandecic suggested would be awesome to get started [18:10:13] Lydia_WMDE: which thing? i'm open to anything, just picked the first one that caught my eye on that list [18:10:22] hehe cool [18:10:23] https://phabricator.wikimedia.org/T92759 [18:10:34] and as far as file size goes, why not just compress the data and let the user's browser decompress [18:10:53] store metrics for searching alongside the compressed data [18:11:12] stoopkid, Lydia_WMDE: unlike other values which we store inline, it would probably make sense to store and version geoshapes separately. Possibly as wiki pages, but could also be something completely different. Or we could just reference an external repo that already has them. [18:11:27] DanielK_WMDE_: yeah [18:11:33] stoopkid, Lydia_WMDE: storing geoshapes inline sounds like asking for trouble. [18:11:39] agreed [18:11:49] but i wasn't thinking of this before it was mentioned in the ticket [18:11:51] ;-) [18:12:04] asking for trouble how, in terms of the file size? [18:12:17] in terms of being able to open the item in reasonable time [18:12:25] and not killing your browser [18:12:29] stoopkid: it has to somehow fir into how we store and version entity data (Q-Items and properties) [18:12:32] *fit [18:13:32] one completely different way to go about it could be allowing geoshape files to be uploaded on commons and link them. [18:13:38] stoopkid: all information about an item is stored in a single json blob. all of it is copied when a single part of the info changes. that's ok if the blob doesn't get to big. If we put the geodata itself into the item's json blob, it will very likely get too big. [18:13:47] Lydia_WMDE: DanielK_WMDE_: well, if the page was loaded asynchronously or sectioned off into multiple pages this would solve 90% of the issue with page size, and then as far as geoshapes go, these could theoretically be provided as a link to separate data [18:13:52] Lydia_WMDE: that'S actually a pretty good idea. [18:14:09] using MediaWIkis "uploaded media" feature to handle geshape files would help [18:14:35] DanielK_WMDE_: copied, you mean to the rest of the wiki sites? [18:14:42] stoopkid: do you want to rewrite the infrastructure for storage, diffing, display, caching, versioning, import/export, etc? [18:14:52] probably not :D [18:15:00] if we go the way of doing this via commons we'll need to have a discussion about it and see if this is acceptable [18:15:03] so, you'll have to live with how wiki pages work [18:15:05] for commons [18:15:59] DanielK please no :) not convinced uploading to commons and treating as files is good idea [18:16:01] i don't think it *has* to be on commons [18:16:02] stoopkid: much less issues should come with the new formula datatype. if you're into adding a new datatype [18:16:12] audephone: better than storing inline. [18:16:25] audephone: there are other options, but they are a lot more work [18:16:36] I would consider to tie into sort of thing like the graph extension [18:16:57] audephone: so you'd want to make the geoshaps a new content model for wiki pages? [18:16:59] stoopkid: that'd be https://phabricator.wikimedia.org/T67397 [18:17:10] Link to that or other external Thing [18:17:16] would work. it's quite a bit of work though. [18:17:28] DanielK_WMDE_: oh wait, so you're saying when you edit a page on wikidata, the entire page is transformed into a json blob and copied back to the server for storage, and when you want to pull that page up, the entire json blob is loaded and copied back to the user? [18:17:35] We have attached kml now [18:17:47] stoopkid: exactly. sounds silly, but has a lot of advantages [18:18:14] what are the advantages to this method, besides simplicity? [18:18:23] (03PS11) 10Smalyshev: T94224: Refactor prefixes to be output only once, eliminate need to clean prefixes on dump [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/199543 [18:18:35] stoopkid: well, not "exactly" - the full JSON blob isn't send back and forht between client and server. But between the app layer and the storage layer on the server. [18:18:42] DanielK_WMDE_: I've removed the drain state, seems to be ok without it do far [18:18:45] Lydia_WMDE: that looks pretty cool as well [18:18:46] *so far [18:19:06] stoopkid: cool :) [18:19:24] stoopkid: if you want to go with that that'd be really awesome and would make people happy [18:19:29] stoopkid: robustness, efficient access for diffs, simple redo/undo of full revisions, simple import/export of full pages. [18:19:39] stoopkid: and: we get all of that for free, since it already exists for wikitext [18:19:46] data size [18:19:57] i'd count data size as a disadvantage :) [18:20:11] to display a thumbnail, the shape might be greatly simplified and smoothed [18:20:25] depending on zoom level etc [18:20:33] sure [18:20:37] DanielK_WMDE_: so really if you could solve this data-size issue that would be a major critical step forward? [18:20:46] And stored just as link [18:21:20] stoopkid: the way to solve it is to store references, instead of inlining the data. which is what I sais initially. [18:21:24] * [18:21:30] bah, can't type today [18:21:36] * DanielK_WMDE_ goes back to writing test cases for rdf export [18:21:49] SMalyshev: yay! [18:21:55] stoopkid: planets for coordinates! :D [18:22:01] that's a really neat task [18:22:32] ah, so users can help the data-size issues just by shortening the pages by separating things out into other objects and then referencing them [18:23:03] stoopkid: yes but i'd rather not push towards this but instead improve performance [18:23:30] both sound good to me [18:23:39] splitting pages is also the way Wikipedia works [18:23:49] when something gets too big [18:24:02] yes but it makes re-use and overview worse [18:24:11] so only as a last resort imho [18:24:41] For a dump or export we can pull shapes in [18:24:48] how does separating things off into other objects and then referencing them make re-use worse? it seems like it's maximizing reusability [18:25:12] take part of a large object, and turn it into a small object that can be referenced by many objects [18:25:29] it's not about reuse inside wikidata [18:25:31] but outside [18:25:36] let's start with a simpler task :) [18:25:39] it's convenient to have all data about once thing in one place [18:25:44] (03PS10) 10Ebrahim: Better handling of bidi content on entity terms table [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/198228 [18:25:48] agreed dennyvrandecic ;-) [18:25:49] before and after on time values, that would be awesome [18:25:59] planets and moons on coordinates, yeah! [18:26:00] Lydia not sure I agree [18:26:22] if I am only interested in geo stuff etc. [18:27:24] copying references easily [18:27:30] (or, at all) [18:27:36] sorting statements [18:27:48] expanding the language labels box [18:28:30] imho the best things to start with are expanding the in other languages box and the formula datatype [18:28:38] they're both rather isolated and very useful [18:28:42] by the way, the transliteration feature on the sitelinks is really neat, whoever did that [18:29:05] and no-one is currently working on them [18:29:13] so it'll not get in anyone's way [18:29:31] the formula data-type is probably what i'll start looking at [18:29:43] \o/ [18:29:58] there are too many issues to think about in geospatial coordinate systems at this point [18:30:01] if you run into issues please do ask here and if no-one is around in the ticket [18:30:14] will do :) [18:30:59] now, question, is there a concrete notion of "types/subtypes" on wikidata? [18:31:20] i mean, can the users extend the type-hierarchy [18:32:05] stoopkid: re coordinates, it's just about extending the UI so that the use can switch to other bodies. the backend etc. is already ready for that [18:32:22] stoopkid: that woul be a pure UI level work, without having to change anything server-side [18:32:37] stoopkid: formulas will be quite harder [18:33:14] git pull from gerrit is unresponsive for me right now, although mtr shows no issues [18:33:28] Just me? [18:33:46] dennyvrandecic: well ill take a look at that [18:35:28] (03CR) 10Ebrahim: "Rebased and no longer DOM operation at least." [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/198228 (owner: 10Ebrahim) [18:35:38] stoopkid: great. I'd really recommend something simple like that. And it ties in into future work on datatypes anyway [18:35:51] ok, so i see the wikidata page for Barack Obama, and it says Barack Obama is an "instance of" the object Human [18:36:06] i follow go to the page for Human, and it says Human is a subclass of Homo [18:36:52] is there a mechanism, "create object of type *human*" [18:37:06] based on the information stored on the wikidata page for human [18:37:23] no [18:37:35] wikidata doesn't have a built in notion of classes and so on [18:37:45] intentionally [18:37:56] how come? [18:38:06] Ok, network problems again... [18:38:13] * hoo keeps calm and uses ProxyCommand [18:39:16] stoopkid: http://blog.wikimedia.de/2013/02/22/restricting-the-world/ [18:39:51] a mechanism for creating subtypes of objects to extend them would make the system quite reusable [18:42:06] it would have advantages but it is a concious decision not to do it for reasons lined out in the blog post dennyvrandecic linked [18:44:19] https://gerrit.wikimedia.org/r/#/c/198228/9/lib/resources/jquery.wikibase/jquery.wikibase.descriptionview.js [18:44:19] "I'd want them injected instead of referring global scope." [18:44:19] Anyone can help on understanding it? I think I have resolved it anyway but like to know anyone around can help me on finding what it means [18:45:07] (03CR) 10Hoo man: [C: 032] Introduce getHelpMessage [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197282 (owner: 10Adrian Lang) [18:45:14] how does it make any less sense to say that "the country of Paris is goat cheese" than "Julius Caesar has a capital", as if he were a country? [18:45:34] or i should ask, why is the 2nd one better than the 1st [18:45:54] (for your system) [18:48:00] actually a typing/subtyping system doesn't need to have any of the "strong restrictions" that are spoken of in that article [18:48:16] while simultaneously encoding all the basic restrictions we do actually want [18:51:50] stoopkid: so, what does a typing system mean for you? [18:52:35] (03Merged) 10jenkins-bot: Introduce getHelpMessage [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197282 (owner: 10Adrian Lang) [18:53:25] "deferreds"... somehow that variable name bothers me [18:53:34] dennyvrandecic: lemme write it up real quick [18:53:42] but I can't think of anything better for a list of deferred [19:09:29] dennyvrandecic: here's a description of my type-system as well as some comments on why i think type-systems don't have to be so restrictive as made out to be in that blog: http://mathurl.com/p5nbe53 [19:10:02] (03CR) 10Smalyshev: Split RdfBuilder [WIP] (0324 comments) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [19:11:23] stoopkid: how would "create object of type *human*" be different than "create entity and add statement: instance of *human*" ? [19:11:25] so Age is a property of Person in your example? [19:11:40] stoopkid: ^ [19:11:48] dennyvrandecic: yes [19:12:07] so, an animal could not have an age? [19:12:11] stoopkid: ^ [19:12:28] jzerebecki: when you assign that Barack Obama is an "instance of" human, does it pull down all the properties of human into Barack Obama? [19:13:06] stoopkid: no, it doesn't. instance of is just a property like any other too [19:13:45] dennyvrandecic: well, we could create another object Animal, assign it the property "Age":*Number*, and then recast the object person as Person{type: *Animal*}, and then the whole system would work the same way but have a wider range of applicability [19:14:08] what about trees? [19:14:42] so we create another object Life-form [19:14:53] then you'll ask what about rocks [19:14:58] exactly [19:14:59] then what about the Earth [19:15:02] then what about the universe [19:15:03] aye [19:15:09] (03CR) 10Smalyshev: Split RdfBuilder [WIP] (031 comment) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [19:15:27] so what is the common type of the universe, barack obama, and the earth? [19:15:47] "thing that have an age"? [19:15:58] dennyvrandecic: probably the concept of an "Event" [19:16:06] dennyvrandecic: http://mathurl.com/o7dejds [19:16:09] so barack obama is an event? [19:16:10] it's a Denny! [19:16:41] dennyvrandecic: after reading this second link you might think yes [19:16:53] well, no [19:17:01] you are suddenly creating a top level ontology [19:17:05] that we first have to agree with [19:17:07] Lydia_WMDE: whoever manages the G+ page of Wikidata put the link to the office hour logs two times (in one post) [19:17:11] dennyvrandecic: of course [19:17:23] lazowik: one should go to the full log and one to the summary [19:17:25] and that is me ;-) [19:17:49] stoopkid: of course? so you want the project first to agree on a top level ontology before we can state that barack obama has an age? [19:17:55] hmm, right [19:18:06] they're just similiar [19:18:10] stoopkid: how is that more effective than the current approach? [19:18:12] * Lydia_WMDE finally leaves the office. back online in 40 mins or so [19:18:58] dennyvrandecic: the reusability, the checks for consistency, the ease of creating and extending data [19:19:18] but we have reusability [19:19:24] we have checks for consistency [19:19:48] and it seems it is easier to create data without types, at least, we don't have to have a commonly agreed top-level ontology first [19:20:51] (03CR) 10Smalyshev: [C: 04-1] "See my detailed comment above. In general, I'm not a big fan so far - now to understand what is going on you need to read 10 files instead" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [19:21:48] DanielK_WMDE_: are you still around? [19:21:51] stoopkid: let's stay simpler. so you say only 4d manifolds have age [19:22:05] dennyvrandecic: why do we need to agree on a top-level ontology? why not just simply record up to the height in the ontology that we actually agree upon, and not try to generalize further? [19:22:19] well, how high do you need to go for age? [19:22:24] dennyvrandecic: just because the system allows for generalization doesn't mean we have to, just like we don't have to say that the capital of Julius Caesar is goat cheese [19:22:44] but we have to, because we need to put the "age" property on a specific type [19:23:51] dennyvrandecic: age is a a simple example for understanding the concept, a bad example for analyzing in detail because we would actually just store a birth-date instead and "age" = (current date | death date) - birth-date [19:24:08] no, no, let's stay with age, i like that example [19:24:40] what about the bible? does it also have an age? [19:24:52] dennyvrandecic: which bible? :) [19:25:06] are we talking about a specific bible, a specific version of the bible, or the entire concept of the bible? [19:25:15] (03CR) 10Hoo man: [C: 032] "I especially like the tests" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197281 (owner: 10Adrian Lang) [19:25:20] https://www.wikidata.org/wiki/Q1845 [19:25:26] this bible [19:26:06] that's a really short page, i would've thought there'd be more information about the bible in there [19:26:41] sure, it could have an age [19:27:03] how does that fit into the type as a 4d manifold? [19:27:20] what do you mean? [19:27:31] with your definition of the type that holds age [19:27:42] so, is the bible an event too? [19:28:00] yes [19:28:01] (03CR) 10Hoo man: "Also not used in any user scripts (only in https://www.wikidata.org/wiki/User:Tpt/validator.js but that looks old an unmaintained)." [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197290 (https://phabricator.wikimedia.org/T88428) (owner: 10Adrian Lang) [19:28:10] ok, what is not an event? [19:28:25] abstract concepts like numbers [19:28:32] (03PS1) 10JanZerebecki: Do not set $wgResourceLoaderMaxQueryLength for tests [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200937 (https://phabricator.wikimedia.org/T90453) [19:28:43] so godels theorem is not an event? [19:28:56] no [19:29:16] so godels theorem would not have an age [19:29:18] (03CR) 10Hoo man: [C: 032] Do not set $wgResourceLoaderMaxQueryLength for tests [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200937 (https://phabricator.wikimedia.org/T90453) (owner: 10JanZerebecki) [19:29:34] it could [19:29:45] it can't, because age only works on events [19:29:45] you can assign Age to anything, even if it doesn't make sense (which in this case it does) [19:29:59] hmm. then i didnt understand your definition of types [19:30:52] we don't need to keep abstracting further and further out until we get to Event [19:31:10] there could be multiple "top-level roots" for the property Age to be inherited from [19:31:23] or there could be a single one [19:31:24] (03Merged) 10jenkins-bot: Split off CachingEntityStore from ApiEntityStore [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197281 (owner: 10Adrian Lang) [19:31:41] depends how much the world agrees on ontology, i guess [19:32:11] the world disagreeing on some very abstract pieces of ontology doesn't break 99% of the system though [19:32:22] we all agree on most of it [19:33:12] so what you are saying is, properties can be attached to any number of types [19:33:40] can an instance have several types or not? [19:33:53] yea a particular property can be attached to any *object*. objects can serve as "types" for other objects [19:34:03] yes [19:34:04] (03Merged) 10jenkins-bot: Do not set $wgResourceLoaderMaxQueryLength for tests [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200937 (https://phabricator.wikimedia.org/T90453) (owner: 10JanZerebecki) [19:34:17] a subtype can have several super-types [19:34:17] ok, and can a type have an arbitrary number of supertypes? [19:34:22] ok [19:35:01] but we could always encode some basic logical rules, like no circularity in the type/subtype definitions [19:35:16] sure, we can do that now too [19:36:03] so, what do I win by using this type system? [19:36:47] let's we have our objects Number (plus it's subtype variants), Person, Barack Obama [19:37:06] can you rephrase? [19:37:27] i'm just listing the objects we've currently been using [19:37:50] i will be back in an hour or so, need to run [19:37:58] ok [19:38:26] anywhere you see "Number" in the system, you could substitute Number with one of it's subtypes [19:38:36] (03CR) 10Daniel Kinzler: Split RdfBuilder [WIP] (0325 comments) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [19:38:38] likewise, anywhere you see "Person" in the system, you could substitute it with Barack Obama [19:39:03] and, because Barack Obama is a subtype of Person, then the properties of Person force consistency checks on the values you're setting for Barack Obama [19:39:47] this "instance of" property wouldn't just be a property, it would cause the object to inherit the properties from the object which it is an "instance of" [19:41:04] stoopkid: AFIK nothing in the current design prevents inheritance/inference [19:41:40] perhaps nothing prevents it, but nothing is making it happen either? [19:42:43] (03CR) 10Daniel Kinzler: "@Smalyshev: Duplication exists, but is minimal. The performance issue needs to be watched, of course." [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [19:44:07] stoopkid: you could model them with instance/subclass of. but much more important things to implement IMHO, so nobody is working on it. [19:44:27] stoopkid: we intentionally opted against the software knowing anything about the relationships between items, and against having fixed schemas/classes, or inheritance. It's a powerful tool, but *very* hard to make work when applied to the complexity of the real world. [19:45:01] though there was doing this in the query service being worked on currently for performance reasons for some selected things [19:45:02] stoopkid: dbpedia is working on a "projection" of some parts of wikidata. that projection would have the "strong" class/instance relationships, but it only works for a small subset of the data [19:45:05] reality is squishy... [19:45:25] thats why we need squishier data structures [19:45:26] err: there was talk of doing this [19:45:35] (03CR) 10Smalyshev: "I don't think the intent here is clarified, rather the opposite - until you read all 10 files, you do not know anymore how RDF is generate" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [19:46:22] (03PS3) 10Hoo man: Remove wbUsedEntities [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197290 (https://phabricator.wikimedia.org/T88428) (owner: 10Adrian Lang) [19:46:58] (03CR) 10Hoo man: "Rebased (cherry-picked) on top of master." [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197290 (https://phabricator.wikimedia.org/T88428) (owner: 10Adrian Lang) [19:47:19] jzerebecki: DanielK_WMDE_: http://www.wikidata.org/wiki/Q19720145 just made this, and assigned "instance of" = "human" [19:47:36] why not just pull in the properties for "human" [19:47:53] and allow the user to substitute these and add more as necessary [19:48:22] stoopkid: primarily because other things like units are more important [19:48:45] jzerebecki: units as in.. physical measurement units? that kind of thing? [19:48:50] yes [19:49:07] (03CR) 10Daniel Kinzler: "Well, each of us understands their own code better. That's natural. I suggest to invite others to be the judge of what is easier to unders" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [19:49:30] that's fair enough, it's simply not a real priority because there are other very important things which need to go into the system? [19:50:03] yes, that is my personal opinion [19:50:45] stoopkid: and notice that you can still query the inherited properties by just following instance/subclass along [19:51:31] even when there is no support for materializing them by wikibase/wikidata [19:51:39] what do you mean by query? just pull up the page for human and check what properties it has and then go back to my person and assign these properties one by one? [19:53:34] seems time-consuming relative to the option of just pulling them into your object when you assign "instance of" [19:54:06] (03CR) 10Hoo man: [C: 032] "Manually verified, works great. On Q183 (Germany) this will cut down the response size by about 6%. On some smaller test items I had aroun" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197290 (https://phabricator.wikimedia.org/T88428) (owner: 10Adrian Lang) [19:54:39] stoopkid: say you want to not only have a list of all entities that have instance of human but also those that have instance of a subclass of human with abitrary depth connection by subclass, then a database that has the wikidata data can give you the answer by following these links [19:54:42] jzerebecki: another way to do it is to only show the properties you've actually specified when you're in "View Mode" and then pull in properties of the super-types when you go into "Edit Mode" so that you have something to reference [19:55:46] stoopkid: the constraints are designed to be able to do that [19:56:52] jzerebecki: they're designed to be able to query a list like that? [19:57:49] stoopkid: no designed to tell you, this is instance of human and a human should have sex-or-gender, but this one doesn't yet [19:58:38] (03Merged) 10jenkins-bot: Remove wbUsedEntities [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197290 (https://phabricator.wikimedia.org/T88428) (owner: 10Adrian Lang) [19:58:50] jzerebecki: oh ok, that's pretty much the majority of the functionality i was talking about, where do i find it where it shows the fields i'm missing for my person: http://www.wikidata.org/wiki/Q19720145 [19:58:58] stoopkid: the query service is designed so it will be able to give you the list [19:59:55] jzerebecki: Is your connectivity to eqiad also degraded right now? [20:00:19] I have less than 10KiB/s to phabricator apparently [20:02:02] :S [20:02:03] stoopkid: the on entity-page constrains are not done yet, a group of students is working on that. on wikidata, currently the constraints are currently being moved from templates on the discussion page of the propert to statements on properties. there is a bot that creates reports, but i'm not update on that. [20:03:11] jzerebecki: oh okay, so parts of this are actually already in the process of being implemented, that's cool [20:03:49] (03Abandoned) 10Hoo man: Merge branches 'getHelp' and 'caching' into lessUsedEntities [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/197289 (owner: 10Adrian Lang) [20:04:45] hoo: https://tools.wmflabs.org/snapshots/builds/mediawiki-core/mediawiki-snapshot-master-ba6431c.tar.gz gives me 3MB/s and rising [20:06:12] jzerebecki: 5.09KB/s [20:06:18] Are you in the office? [20:06:20] yes [20:06:42] 2.88KB/s yay [20:07:07] dtag? [20:07:14] 1and1 [20:07:37] network wise the same thing, right? [20:07:43] Probably [20:08:16] I could VPN, but I'd prefer not to as that's not very punctual :P [20:08:28] sjoerddebruin: pffffff, deduped 100+ rkd links :-) [20:08:34] <3 [20:09:27] * aude waves [20:09:53] aude: Just of of curiosity: How's your connectivity to eqiad right now? [20:09:58] 1.13KB/s [20:10:04] let's see [20:10:10] wget https://tools.wmflabs.org/snapshots/builds/mediawiki-core/mediawiki-snapshot-master-ba6431c.tar.gz [20:10:38] hoo: 1MB/sec? [20:11:31] it's unusable :( [20:11:52] Yep :( [20:11:55] Interesting. Can you pastebin a traceroute / mtr ? [20:12:15] I'm taking the tele2 route [20:12:52] https://phabricator.wikimedia.org/P463 [20:12:54] 330KB/s from my German vserver [20:13:05] * aude using usb stick [20:13:09] but normally not this slow [20:14:27] I'm not going via telia [20:14:45] telefonica [20:14:51] You have some funky async routing aude [20:15:51] with proxycommand, gerrit fast [20:15:52] Off to some food... hopefully this improves later on... terrible right now :S [20:21:28] JohnFLewis: see admin chan [20:21:52] kay [20:24:35] (03PS4) 10Daniel Kinzler: Split RdfBuilder [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 [20:24:43] (03CR) 10jenkins-bot: [V: 04-1] Split RdfBuilder [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [20:36:47] (03PS5) 10Daniel Kinzler: Split RdfBuilder [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 [20:37:23] (03CR) 10Daniel Kinzler: "PS5 is a rebase." [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [20:38:39] (03PS6) 10Daniel Kinzler: Split RdfBuilder [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 [20:42:57] (03CR) 10jenkins-bot: [V: 04-1] Split RdfBuilder [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [20:43:11] stoopkid: also, if you add a property, the editor will suggest you what to add anyway [20:43:16] based on frequency [20:43:23] that is quite some nice work, done last year by students [20:44:38] dennyvrandecic: that is a nice feature, we were also talking about a feature kind of *similar* to what i'm talking about that students are currently implementing [20:44:56] idk if you know anything about that in particular [20:45:32] (03CR) 10Daniel Kinzler: [C: 032] T94224: Refactor prefixes to be output only once, eliminate need to clean prefixes on dump [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/199543 (owner: 10Smalyshev) [20:45:51] yes, checking for restrictions [20:45:55] DanielK_WMDE_: yay \o/ [20:46:10] but all of this does not require a type system [20:46:22] restrictions can be much more expressive than a type system can capture [20:47:31] (03Merged) 10jenkins-bot: T94224: Refactor prefixes to be output only once, eliminate need to clean prefixes on dump [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/199543 (owner: 10Smalyshev) [20:49:10] (03PS7) 10Daniel Kinzler: Split RdfBuilder [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 [20:49:19] (03CR) 10jenkins-bot: [V: 04-1] Split RdfBuilder [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 (owner: 10Daniel Kinzler) [20:50:37] SMalyshev: I'll rebase https://gerrit.wikimedia.org/r/#/c/200183/ now. Are you ok to merge it? [20:50:50] orr, the split thing is still failing. [20:51:47] (03PS5) 10Daniel Kinzler: Moving RdfWriter and friends to a separate component called Purtle. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200183 (https://phabricator.wikimedia.org/T94172) [20:52:09] will probably conflict with the prefixes patch now, though [20:56:11] (03PS8) 10Daniel Kinzler: Split RdfBuilder [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/200494 [20:58:18] hey there. if i wanted to get a list of all states (I guess Q7275) out of WikiData, how would I do that? can that be done on the website, or just through the API? [20:58:18]