[00:20:08] Hi, how do I do reverse search in SparQL in wikidata? Like: For every "platform" that "libreboot" runs on, select "label" and "mass" ? [00:20:21] Reverse or combined queries, sorry [00:22:46] I've tried to do it and I ended up with that: [00:23:02] https://en.wikipedia.org/w/index.php?title=User:GNUtoo/sparql&action=edit [00:23:16] Note that adfeno is trying to help me with making SPARQL querries [00:24:01] adfeno: can you see the SPARQL in that page? [00:24:20] Note that in the non-edit page it doesn't show up in the querry tool [00:25:22] What I didn't manage to do is filter out "ARM" for instance [00:25:53] like if the platform is a CPU architecture => do not select it, or only select if it's an instance of laptop [00:26:29] And note that I did the query by looking at examples and modifying them without really understanding what I was doing [00:26:40] GNUtoo: Yes, I can see the SparQL query in the edit mode. [00:26:56] I get the OPTIONAL keyword but the gas:service is totally obscure to me [00:27:29] * GNUtoo needs to find a good tutorial on SPARQL [00:28:16] * GNUtoo knows SQL [00:28:37] GNUtoo: I have this open, I stared to read it https://www.wikidata.org/wiki/Wikidata:SPARQL_tutorial#Our_first_query [00:28:56] Thanks [00:29:15] I too know sql and am trying to do a few things, and have done 1! [00:46:06] GNUtoo: I think I might have found something [00:47:22] I'll do some testing [00:48:31] is there a style guide for capitalization, like I am surprised this is all lowercase: urgent care clinic: type of medical clinic http://www.wikidata.org/entity/Q7900484 [01:33:42] How do I iterate through all "platform" in "Libreboot"? Tried this, but didn't work: … [01:34:21] … SELECT ?device ?deviceLabel ?subclassLabel ?dateLabel [01:34:21] WHERE { ?device wd:Q20085696 wdt:P400. ?device wdt:P279 ?subclass; wdt:P577 ?date. SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE]". } } [01:36:11] Put ?device in the end before first dot, but no use either. [01:38:04] If I keep only the part up until the first dot it works [01:38:37] But I also want to get the "platfom" image too. [01:39:57] Or subclass [02:54:16] How to select wd:Example1 or wd:Example1 both from ps:ExampleA ? [02:57:03] I succesfully did it for wd:Example1, but using "wd:Example1, wd:Example2" or "|" gives error. [03:05:27] Found it, use FILTER( ?var in ( opt1, opt2 ) ) [08:05:19] Because P856 now asks what language the website is in, I added a few for (the contents of) this digital library. I'm not sure this is useful or correct, though. https://www.wikidata.org/w/index.php?title=Q3639582&diff=1169815743&oldid=1169810090 [09:33:48] has the font changed for the diffs? [09:34:24] nikki: yes, it was announced on last Tech/News: https://phabricator.wikimedia.org/T250393 [09:35:26] I see... [09:39:31] Nemo_bis: how do you switch to those languages? I only see hebrew in the language menu [09:42:05] nikki: they're the languages of the content, not hte interface [09:42:50] if I'm not mistaken, people usually use those properties to mean the interface, especially for sites where the content can be essentially any language [09:43:10] I couldn't find any content though, dunno if I just don't know what to search for :D [09:43:29] Hm. But the interface is not even determined by the website, it may change with the next software update [09:43:38] And should we then list 400 languages for every MediaWiki website? [09:44:09] Well yes this proprietary software is terrifying. The result of decades of monopolies [09:44:29] But see for instance: [09:47:47] there's always things like https://www.wikidata.org/wiki/Q20923490 (not sure whether I like it but...) [09:48:13] anyway that was just how I interpreted it, maybe it's not actually what other people do, you'd need to ask on wiki probably [10:13:02] I want these hospitals added - is the phone and license # appropriate? https://docs.google.com/spreadsheets/d/1uJ5r1qlXl5YLXvMHREDKtsfULy-P3xfU [11:20:12] please voat https://www.wikidata.org/wiki/Wikidata:Property_proposal/Goodreads_work_ID [11:55:35] CarlFK: I think so [11:55:46] just have to make the numbers international [11:55:48] but yes [11:55:49] it is [11:56:33] CustosLimen: so add a +1 for us in front ? [11:56:44] CarlFK: https://www.wikidata.org/wiki/Property:P1329 [11:56:45] yes [11:57:00] e.g. +1-202-456-1414 [11:57:03] bleh, VIAF is one of the most shitty datasets :D [11:57:15] thanks [11:57:20] are there even humans at work? [11:57:20] (217) 223-1200 -> +1-217-223-1200 [11:57:32] SothoTalKer: what are you trying to do with it? [11:57:50] CarlFK: and make sure to use right properties [11:57:52] and not make dupes [11:58:39] avoiding dups is one of my 'projects' [11:59:37] it lead me to using Open Street Map to get lat/lng which I can then query wikidata for "anything within 500 meters, double check to see if it is the same thing [11:59:49] okay [12:00:10] which lead me to: Wow, the map editor for OSM is nice. [12:03:45] not too much [12:27:38] I'm trying to link the two wiki articles https://en.wikipedia.org/wiki/Lumbar and https://de.wikipedia.org/wiki/Lendenwirbels%C3%A4ule but on wikidata I'm getting "Could not save due to an error." when I hit publish [12:28:02] wikidata page: https://www.wikidata.org/wiki/Q1293405#sitelinks-wikipedia [12:29:17] I was able to merge the items [12:29:54] not sure why you got an error, I thought maybe it was the bug I've seen a few times recently but apparently not [12:32:41] well ok then :-/ [12:48:13] Hi there, does WikiData expose a HTTP location with which one can send query and have the website returm, for example, a .CSV, .JSON, .YAML ? [12:48:30] or a Recutils .REC format? [12:50:23] adfeno: see https://www.wikidata.org/wiki/Wikidata:Data_access [12:56:46] I wonder which exact headers or keys does https://www.wikidata.org/wiki/Wikidata:Data_access#SPARQL_endpoints expect ? [13:09:52] Hm, direct access to https://query.wikidata.org/sparql doesn't make it clear either :S [13:25:43] OK, found the way, it's in the SPARQL protocol specification ( https://www.w3.org/TR/sparql11-protocol/ ). :D [13:30:11] adfeno: sorry I to sleep, I'm back now [13:30:39] Oh my, sorry for waking you up [13:36:31] I found Wikidata [[Property:400]] very misleading when describing [[Q20085696]]. [13:36:33] 10[1] 04https://www.wikidata.org/wiki/Property:40013 => [13:36:35] 10[2] 10https://www.wikidata.org/wiki/Q20085696 [13:36:54] The last one is Libreboot [13:37:18] Reason being that computer models aren't really platforms, but computer models :D [14:01:31] SothoTalKer: I just ask because there are other things to use [14:01:35] that maybe suck less [14:01:37] for specific domains [14:02:12] adfeno: I've the same issue [14:02:32] I'd like to have something like "compatible with" [14:02:36] or "supports" [14:02:49] Here, the idea is to bind software and hardware in some way [14:03:09] Example: exynos4412-i9300.dts could support the GT-I9300 [14:03:11] There is nothing like that already? [14:03:22] I didn't find it [14:03:38] I probably missed it though [14:04:30] CustosLimen: Also, Wikidata [[Property:400]] ("platform") is very misleading when describing [[Q20085696]] ("Libreboot") since it includes actual computer models instead of just architectures. [14:04:31] 10[1] 04https://www.wikidata.org/wiki/Property:40013 => [14:04:34] 10[2] 10https://www.wikidata.org/wiki/Q20085696 [14:04:45] yes [14:04:53] okay [14:04:56] bit of a mess [14:04:56] Well, it probably needs to have both informations [14:04:59] Besides, even the concept of "platform" is confusing by itself, not very specific. [14:05:02] god luck [14:05:03] The architecture supported [14:05:14] and the devices it support as well [14:05:22] which are two completely different things in that context [14:05:31] ]] [14:05:39] [[Property:P2360]] maybe should be changed [14:05:40] 10[3] 10https://www.wikidata.org/wiki/Property:P2360 [14:05:45] it makes no sense as it stands [14:06:03] Ideally having a more fine grained way to describe what it supports or what is supported by it would be better [14:06:20] For instance it supports computers with the GM45 chipset and has code for the RAM init for the GM45 [14:06:35] so the GM45 chipset could be supported by Libreboot and Coreboot for instance [14:06:42] pls vote: https://www.wikidata.org/wiki/Wikidata:Property_proposal/Goodreads_work_ID [14:07:44] * GNUtoo doesn't know what Goodread is beside what is written on the wikidata page [14:07:50] Though I'll look at it [14:08:38] Interesting [14:09:03] CustosLimen: do you work on inventaire.io ? [14:09:21] what is that [14:09:27] first I hear about it [14:09:37] ah yes [14:09:39] It's a website that uses wikidata [14:09:40] I have seen it [14:09:47] It's for sharing books or something like that [14:09:56] yeah I registered [14:09:58] What's interesting is that they have a presentation about it and documented the model [14:09:59] I don't work on it no [14:10:17] So I'm very interested in it as I want to reuse wikidata data in various wikis [14:12:34] CustosLimen: There is something I don't understand with the Goodreads, [14:12:39] When a company have multiple sub companies with different brands, should the sub company use owned by or parent organization? [14:13:03] You use the "work ID" as description for it, but it only points to work/editions/$1 [14:13:10] yotam: parent organization [14:13:12] I think [14:13:15] And edition is a work? [14:15:15] GNUtoo: well, depends on what you mean by edition or work. A goodreads edition is similar to a wikipedia edition yes, I chose that URI because if you go to https://www.goodreads.com/work/4912783 then it redirects to an arbitrary edition and does not show you any info about the work itself [14:15:42] But really the id here is a goodreads work ID, not an edition ID - https://www.goodreads.com/work/editions/4912783 [14:16:22] Can you get the work id from a precise book edition? [14:16:23] Edition IDs (Book IDs) are resolved via: https://www.goodreads.com/book/show/4912783 [14:16:31] GNUtoo: yes [14:16:42] How do you do that? [14:16:46] if I go to a book there is an all editions link [14:16:54] CustosLimen: What exactly is the difference? I see that Shazam for example is owned by Apple [14:16:54] ok [14:17:11] yotam: not sure actually [14:17:17] I use parent organization [14:17:18] https://www.goodreads.com/book/show/658332.Free_as_in_Freedom [14:17:20] I don't find it [14:17:29] *I didn't find it [14:17:45] https://www.goodreads.com/book/show/658332.Free_as_in_Freedom There is a link there "All editions" with https://www.goodreads.com/work/editions/644416-free-as-in-freedom-richard-stallman-s-crusade-for-free-software [14:18:02] so 644416 is the work ID [14:18:07] ahh it just didn't show up without javascript [14:18:17] nasty [14:18:49] it shows up here: curl 'https://www.goodreads.com/book/show/658332.Free_as_in_Freedom' | grep 644416 [14:18:52] so it is scrapeable [14:18:54] https://www.goodreads.com/work/editions/644416-free-as-in-freedom-richard-stallman-s-crusade-for-free-software and https://www.goodreads.com/work/editions/57035345-free-as-in-freedom-2-0-richard-stallman-and-the-free-software-revolut are different then [14:19:11] And it shows translations [14:19:27] GNUtoo: yes [14:20:03] So If the content is changed it's a different work [14:20:07] There would also be different items instance of work for "Free as in Freedom" and "Free as in Freedom (2.0)" I think [14:20:15] yeah, somewhat [14:20:22] but if it's translated or published in another format then it's an "edition" [14:20:34] It's a bit missleading as I though that two were different editions [14:20:52] like if the content change, the edition would change, and the rest would be format for me [14:20:52] https://www.wikidata.org/wiki/Wikidata:WikiProject_Books [14:21:10] That explains the data model a bit [14:21:14] But I'm not an expert in books and English is not my native language [14:21:18] so I might have bias [14:21:34] GNUtoo: it is a bit confusing [14:21:42] but that is how wikidata models it [14:21:51] and how other systems model it [14:21:57] oh ok, as long as it's well documented I think it should be ok [14:22:00] > We used the Functional Requirements for Bibliographic Records (FRBR) model (a widely used and famous conceptual framework in library science) to divide the Book properties in 2 sets. The FRBR model underlines the fact that you can view a book in 4 different ways, or levels: as a work, as an expression, a manifestation or an item. [14:24:33] GNUtoo: things are not that cut and dry - the point though is - War and Peace (Q161531) has 100s of different ISBNs [14:24:52] we want to have some way to refer to just the work, and we don't care about the ISBNs [14:25:01] we want some thing to link all those ISBNs back to [14:25:33] CustosLimen: well, that is what wikidata is for [14:25:39] yes [14:25:57] but it would be good to have some ways to both check and populate wikidata [14:26:00] CustosLimen: Using the Goodreads work id for that would not be a good idea as you then depend on Goodread to do the work for you [14:26:12] But having that information looks useful anyway [14:26:19] GNUtoo: why use any external ID then? [14:26:36] Because it's an important information [14:26:42] GNUtoo: I mean the same could be said about ISBN, and OCLC [14:26:47] Exactly [14:26:57] Is OCLC control number important? [14:27:01] how often do you use them? [14:27:04] I don't know OCLC [14:27:10] ISBN is important information too [14:27:30] It is important for cross referencing and validating data [14:27:31] But it's probably not the ISBN which should decide what is what for wikidata [14:27:37] Yes [14:27:46] I'm not saying that the reference is a bad thing at all [14:27:52] It's actually very useful [14:28:13] I'm just saying that it's up to wikidata to define what is a book, not Amazon [14:28:18] all external identifiers ultimately serve to enhance the ability to validate and populate wikidata [14:28:22] indeed [14:28:25] that's my point [14:29:11] but relying only on a single external identifier made by a company to have the ultimate authority on what is a work is probably not a good idea [14:29:20] having as many identifiers as possible is a way better idea for instance [14:54:55] CustosLimen: well, VIAF is the defacto authority regarding people. It aggregates data in a good way already. the problem i have that some data won't be aggregated and others is, but should not :P [14:55:50] annoying [14:56:05] To use it in queries [14:56:10] oops [14:56:18] scratch that I was lines before in irssi [15:01:09] regarding to books ... well, complicated. There is the underlying work, which is just the work itself. Then there are publications. Each publication nowadays gets an ISBN, if its in book form. Then there are editions and translations, the latter basically a derivative work of the original one. Each get their own publications. :) [15:01:51] The Goodreads work ID is for the underlying work [15:03:16] most works will only have 2 items. the work itself and a published book of that work. [15:03:40] yeah, as war and peace does [15:03:50] famous works can have a lot of items [15:03:52] the published works are Bookreads book IDs [15:04:06] yeah war and pace has 4 editions (which correspond to publications) [15:04:13] on wikidata [15:04:18] any edition and translation will be a new item [15:05:06] well, except a second run of the original publication, if its basically identical. [15:05:22] ok done [15:05:36] I've commented [15:06:08] adfeno: CustosLimen: Thanks a lot for the "intended public (P2360)", I've been looking that property for hours and months [15:06:18] *looking for that property [15:06:19] ah [15:06:29] great to help [15:06:46] pretty odd one if you ask me but glad it helps [15:06:53] It's much better than platform, but still not 100% accurate [15:07:05] yes it is not [15:07:09] can I tell that Linux is "intended for" a Lime 2 for instance [15:07:17] or designed for? [15:07:28] well - with how that property is now, not ideally no [15:07:34] but with some changes to property it would make snese [15:07:42] yes indeed [15:08:26] Though you could say that Replicant is intended for people and runs on smartphones [15:08:44] or OpenWRT is intended for WiFi access points and runs on a WNDR3800 [15:09:04] Or better: GuiX is intended for developers and runs on a Lime2 [15:09:31] subject item of this property [15:09:35] target audience [15:09:44] That's probably the key thing I should be looking for [15:10:09] P400 (Platform): [15:10:09] subject item of this property [15:10:13] computing platform [15:10:18] hmmm [15:11:07] * GNUtoo wonders if there is a way to list the possible "subject item of" properties [15:12:16] I've a pretty good idea on how to model hardware from a software background [15:12:29] like I'm mostly interested in software support for given hardware [15:12:41] This means drivers, dts, firmwares, etc [15:14:01] For instance a modem might be supported by many different userspace drivers and a kernel drivers and itself runs GNU/Linux on several cores and run nonfree OS on other cores [15:14:32] or an MMC controller might be supported by Linux, u-boot, Coreboot, barebox, etc [15:15:05] So for that I need to model the hardware [15:15:15] and tell that foo software at bar version support that hardware [15:15:50] So for now I used platform for expressing very very basic things [15:17:05] There are already devices that have multiples computers in them which runs fully free software, so you have different Linux kernels running in parallel on different architectures [15:17:10] Example: The Talos II [15:17:33] So you need support on computer A to talk to computer B which also runs GNU/Linux [15:18:19] So maybe something like "runs on" and "can talk to" could work in an abstract way [15:18:33] Then there is the software components involved [15:20:33] sorry I had irssi hanging [15:20:50] So the issue with that model is then that a DTS doesn't "run" [15:20:55] as it's not code [15:21:05] but I guess that the abstractions would be good enough [15:22:28] Or maybe a more fine grained model would sort that out automatically [15:23:23] Or maybe using a qualifier for peripherals or host or something [15:23:48] or supports + runs on [15:24:01] like you support an HDD but you run on a host computer [15:24:06] and not on the HDD controller [15:36:12] I've found that and nothing matches: https://www.wikidata.org/wiki/Wikidata:List_of_properties/work#Wikidata_property_for_software [15:44:32] At the end I'll probably save time creating a property [18:59:36] GNUtoo: I know the feeling [18:59:50] I have been deliberating for weeks on whether or not to create ETSI identifiers [19:03:02] I'm wondering [19:03:05] with wd cli [19:03:15] or really wikibase-edit [19:04:03] I want to add claims but only if they don't exists [19:06:36] or really even with quickstatements [20:56:18] gnnnnnnnn [20:56:55] why are there property constraints when they are wrong XD [21:12:20] what now? :D [21:16:15] https://www.wikidata.org/w/index.php?title=Property:P1315&diff=1170247969&oldid=1141902825&diffmode=source [21:17:26] i have never seen a valid Trove ID for people and organizations that is longer than 7 digits [21:18:03] trove IDs for works and editions can be longer, but we do not have them in Wikidata [21:18:51] it should likely be added, but i'm not too interested in books, as i cannot read [21:22:59] Example for a Trove work ID: https://trove.nla.gov.au/work/6160665 [21:28:48] anyway, changing the constraint gave almost 250 results where P1315 was filled with values that belong to P409 :p [23:03:22] I've created a new property proposal [23:03:27] but I'm unsure where it went [23:03:43] The name is "Compatible with" [23:05:57] I don't see it in user contributions [23:07:30] https://www.wikidata.org/wiki/Wikidata:Property_proposal/Compatible_with [23:07:48] I've clicked on do it and saved the resulting edit but I see nothing in the page its supposed to have edited [23:08:35] Maybe it just changed this page and added cathegories to be picked up later? [23:23:13] GNUtoo: look what it says at the top of the page :) [23:24:11] So I just need to copy-paste the content I guess [23:24:35] no. you have to transclude it :] [23:25:29] click on the "do it" part, I think it will be self explanatory [23:27:53] and you might want to ping participants of wikiprojects where this property might be useful, like https://www.wikidata.org/wiki/Wikidata:WikiProject_Informatics [23:28:07] Thanks [23:39:41] SothoTalKer: I know undestood, but I'm unsure the definition of the verb transclude is right, https://en.wiktionary.org/wiki/transclude says: [23:39:47] (programming) To substitute a template or other input for its rendered text, such as when parsing wikitext. To include by transclusion. To process fetched data in-line. [23:39:49] on an unrelated personal note: i kinda like old novels, like in 100+ years old [23:40:22] but transclusion https://en.wiktionary.org/wiki/transclusion#English says: [23:40:28] " (computing) The inclusion of part of one hypertext document in another one by means of reference rather than copying." [23:40:50] I understna dthe latter but the former one is not clear to me [23:41:43] the wiktionary one is more abstract and broad [23:42:20] like both definision seem to contradict each other [23:43:02] If I take "transclude (Verb)" I assume it means that you run the pre-processor and gets the text out of it [23:43:23] as the "transclusion (noun)" is inclusion by reference which is something totally different [23:43:28] transclude and transclusion are not valid english words, anyway :P [23:43:34] like the verb describe something like the C preprocessor [23:43:56] whereas the noun describe the fact of using #include to reference something without actually copying it there [23:44:23] ok, as I understand the wiktionary is a dictionary that describes how words are used [23:44:57] So if some words commonly used are not in the so called "official" dictionaries they can still be there [23:45:20] transclude describes here the actual process that happens when a page that has a transclusion in the source. [23:45:32] oh ok [23:45:34] I get it now [23:45:43] Thanks a lot [23:46:06] yw. [23:46:13] Is it only for inclusion or preprocessing / templates in general? [23:46:36] * GNUtoo guesses it's the later [23:47:58] include and transclude are similar, i think, but not identical. [23:48:33] There is also transpiler [23:48:52] which takes as input a language and outputs another language [23:49:01] like valac for instance [23:49:46] Though the resulting C sources doesn't look very human, but I heard that some transpilers can make the resulting source look human [23:50:15] (I heard of a free software one that can transpile from Cobol to Java in a human readable way) [23:51:01] Have the US unemployment Services heard of that? [23:51:07] transclusion is used in wikipedia sites for composite pages like the list of property proposals. it is usually easier/faster to edit small pages, but it is also nice to have an overview of all proposals that exist [23:52:42] also pages that have common sections (which are basically identical) usually transclude a template [23:56:44] The transpiler I heard about was called NACA [23:57:48] (In French: https://linuxfr.org/news/projet-naca-2-transcodage-automatique-vers-java-de-4-millions ) [23:58:17] According to that post the resulting source code can be read by humans [23:58:28] I didn't test that software though so I don't know how much readable it is [23:58:44] now is the question: are modern JIT compiler implementations in emulators just transpilers from one computer code assembly into another one? [23:59:58] The question is what is source code