[00:42:40] About the natural numbers validator, I think it should have an extra condition that it is not an empty string (i.e. that it must contain at least one digit) (re @wmtelegram_bot: I am particularly interested in a discussion about the type proposal system itself, and about the natural numbers ty...) [00:43:27] And here are a couple of suggestions about the type proposal system itself. https://www.wikifunctions.org/wiki/Wikifunctions_talk:Type_proposals (re @wmtelegram_bot: I am particularly interested in a discussion about the type proposal system itself, and about the natural numbers ty...) [01:14:07] about the `|` issue – the API should already be able to handle this for you as far as I’m aware: https://www.mediawiki.org/wiki/API:Data_formats#Multivalue_parameters [01:14:46] the JS code seems to use `mw.Api`, which should even correctly encode arrays automatically (i.e. use U+001F as the separator if any value contains `|`) if you removed the `.join( '|' )` on the parameters [01:14:54] but maybe there’s a reason it doesn’t work out that I’m not aware of [01:25:44] Here's my suggestion for the validator Z13489, with plenty of tests. (re @Toby: About the natural numbers validator, I think it should have an extra condition that it is not an empty string (i.e. that it must...) [01:26:06] re Toby: that sounds good! Added the condition to the validator, and answered on wiki on the suggestions [01:26:13] Oh, that's great, I will link to that [10:32:57] I created https://www.wikidata.org/wiki/Lexeme:L1273381 (unfortunately, I also created L1273382 by mistake 🙄) [10:33:45] L1273381 first 😉 (re @wikilinksbot: L1273382 – natural number [en: English]) [14:51:22] Thanks for expressing your perception, @amire80 . I agree that workflows and UX around Lexemes leave quite a bit space for improvement. I disagree with your dismissal of tools built by the community on top of that. [14:51:23] [14:51:24] But I do have one question: regarding your question for progress bars, I don't understand what that means within the context of a dictionary. When translating the messages for a tool, I understand how to make a progress bar. But for a dictionary? I'm at a loss here. There is no number of words in a language we can refer to. Or what am I missing here? [14:57:57] Can functions on WF query data from Wikidata directly for now? [15:01:16] I love that phrase (re @Jan_ainali: My point is that it isn't a drop in an ocean. It's a drop in a bucket. And you actually don't need that many drops to fill a buc...) [15:01:31] Not yet, it's a future to-do (re @cvictorovich: Can functions on WF query data from Wikidata directly for now?) [15:02:47] It’s important for mmee to (re @vrandecic: Not yet, it's a future to-do) [15:03:41] I would use it to build functions dealing with natural language [15:04:17] It's on our priority list! [15:04:57] Current functions on French conjugation created by me are problematic [15:05:20] Many verbs are irregular [15:05:52] To exclude the irregularities one by one isn’t efficient [15:07:45] Group 2 verbs are of no pattern [15:08:14] Many verbs that seems to be in group 2 actually belongs to group 3 [15:14:57] I don't dismiss the tools in principle. There are tools for advanced Wikipedia things, and while I'd argue that _some_ of them should be built-in (global contributions, pageviews, and some others), it's largely an ok ecosystem: most important things are a part of the platform (that includes extensions), and more advanced things are separate tools. [15:14:57] [15:14:59] In Wikidata, the reliance on tools is much bigger, and in Lexeme, it's almost exclusive. I don't have an exact number, but the feeling is that whenever people (not just me) ask how to do something on Wikipedia or Wikisource, using an external tool is suggested about 10% of the time. On Wikidata, it's 60%, and with Lexemes, it's about 90%. Yet again, that's a feeling, not an exact number. [15:15:00] [15:15:02] What's worse, most of the Lexeme and Wikidata tools have names that I can't remember and that don't describe their action (Tabernacle, Hangor, QuickStatements, Ordia). Their UI is inconsistent, hard to understand and requires training, and many of them are not localized (or localized using Magnus's own tool, which some people somehow manage to use, but I find completely inscrutab [15:15:03] le). (re @vrandecic: Thanks for expressing your perception, @amire80 . I agree that workflows and UX around Lexemes leave quite a bit space for impro...) [15:16:37] Lexème form tool is okay (re @amire80: I don't dismiss the tools in principle. There are tools for advanced Wikipedia things, and while I'd argue that some of them sho...) [15:17:05] I don’t use many of them, and am using some of them [15:18:11] Evidently, those tools are used by some people, who go on making many thousands of edits, and that's probably cool, but I strongly suspect that it's not really a healthy that will reach _masses_. Wikipedia, in comparison, reached many more people. That is its real power and its real, most important asset—the many thousands of editors. I wish it reached even more, but already no [15:18:12] w it has a much wider appeal to potential editors than Wikidata and Lexeme. [15:18:14] [15:18:15] Some people could argue that it's a systemic, structural problem, and that lexicography and structured are nerdy and have a very particular appeal to begin with, but I'm really not sure it's true. With a more consistent UI, they could have a much wider range of people. (re @amire80: I don't dismiss the tools in principle. There are tools for advanced Wikipedia things, and while I [15:18:17] 'd argue that some of them sho...) [15:23:06] The progress bars I fantasize about for Wikidata, Lexeme, and Wikifunctions, are not the same as progress for translatewiki (this is also a reply to @Jan_ainali, who mentioned somewhere that I may be too accustomed to translatewiki). The translatewiki progress are indeed completely straightforward: there are X messages to translate and it's easy to count how many of them are left [15:23:06] to translate. (The total X changes practically every day, but it's still easy to calculate at any given point in time.) [15:23:08] [15:23:09] Wikidata and Lexeme are much more open in that regard, but it's still possible to define some useful progress indicators: (re @vrandecic: Thanks for expressing your perception, @amire80 . I agree that workflows and UX around Lexemes leave quite a bit space for impro...) [15:24:07] 1. On Wikidata, all the property labels, and property labels for particular groups of properties. [15:24:53] 2. On Wikifunctions, all the labels and lexemes that have to be translated for a full localized experience of a given function. [15:25:03] 3. On Wikifunctions, all the labels of a group of functions. [15:27:25] 4. On Lexemes, make a lists of basic nouns that would be expected to be found in any language. (This can have several levels: 100 nouns, 500 nouns, 5000 nouns.) [15:31:33] See the message to which I'm replying here. [15:31:33] [15:31:35] Examples 2 (all the labels and lexemes that have to be translated for a full localized experience of a given function) and 3 (all the labels of a group of functions) are basically what this whole big discussion started from: I'm translating the extension messages, but they depend on labels on wikifunctions.org . (re @amire80: Hallo) [15:35:55] For that message, from which this discussion started, it is not even documented in qqq: https://translatewiki.net/wiki/MediaWiki:Wikilambda-functioncall-metadata-implementation-type/qqq [15:35:56] [15:35:57] So it's not really possible to even localize it. Every day I quietly and happily improve qqq for lots of messages in lots of MediaWiki extensions by figuring out the necessary clarifications myself or by asking the developers or the product managers. This was yet another question of this kind, which turned out to be much more complex than I thought it would be. And I still don't [15:35:59] know how to translate those labels. [15:51:59] (I sometimes complain when people use weird jargon or Q/P/L/Z numbers without explanation, and I try not to do it myself, so I should probably explain what "qqq" is. In case anyone doesn't know, in the jargon of MediaWiki developers and translators, "qqq" refers to message documentation, which explains how a string in software UI is used and gives suggestions about how should it [15:51:59] be translated. If I say anything else that is unclear, never hesitate to ask me.) [15:52:14] I have to mention Italian Wikipedia here [15:52:50] Some of you may know about its custom style sheets [15:53:06] Different colors for different namespaces [15:53:23] But still it wasn’t implemented in Vector 2022 [15:53:39] Valid on classic Vector skin [15:54:17] (Sannita surely knows about it [15:54:36] @Sannita [15:54:57] not sure I can help with that, to be honest [15:55:13] Is it related to Wikifunctions? (re @cvictorovich: I have to mention Italian Wikipedia here) [15:55:16] I can ask around, but I don't see why are we talking about this (re @cvictorovich: Different colors for different namespaces) [15:55:35] No, but related to consistent interface designs [15:56:26] Or I wouldn’t have mentioned it out of nowhere [15:57:55] I'm not a very big fan of custom per-wiki styles and gadgets, but those customizations are minor, and nowhere near the huge differences between Wikidata itself and Wikidata tools, and the huge differences between each of the different tools. (re @cvictorovich: No, but related to consistent interface designs) [15:58:46] Something in new softwares may have broken custom styles [15:58:51] Again, the tools' existence is legitimate and even desirable in general. What bothers me is the outsized dependence of Wikidata and Lexeme on those tools, when compared to Wikipedia. [16:11:00] True there is a lot of irregularities (especially in the most common verbs) but actually most verbs are regular and in the first group (re @cvictorovich: Many verbs are irregular) [16:11:55] Agreed, don't do that. That's what we have Wikidata for. (re @cvictorovich: To exclude the irregularities one by one isn’t efficient) [16:14:41] And if you look only at the infinitive, some are in both 🙃 (re @cvictorovich: Many verbs that seems to be in group 2 actually belongs to group 3) [16:20:33] Nicolas, that’s why I halted the progress on functions for French verbs (re @Nicolas: And if you look only at the infinitive, some are in both 🙃) [16:23:59] I need Wikidata support very very much [16:26:08] Sure, I understand [16:26:09] That said there is still a lot of regular part that should have a function (like all the first group conjugation) (re @cvictorovich: Nicolas, that’s why I halted the progress on functions for French verbs) [16:26:39] Group 1 is usable for now [16:26:42] and vice-versa, Wikidata need Wikifunctions, it goes both ways 😉 (re @cvictorovich: I need Wikidata support very very much) [16:26:55] For group 2, WIP [16:27:09] And group 3 is usable at large [16:27:53] we still don't have a function for each conjugated form tho (re @cvictorovich: Group 1 is usable for now) [16:28:10] Only one exception, “aller” (re @cvictorovich: Group 1 is usable for now) [16:28:16] I’ll do that (re @Nicolas: we still don't have a function for each conjugated form tho) [16:29:12] Lately my request for indefinite functioneer access has passed [16:30:25] By the way, it should be safe to delete [[Z1229]] for now? [16:30:26] https://www.wikifunctions.org/wiki/Wikifunctions:Requests_for_deletions#Z1152_&_Z1229 [16:31:36] Y [16:33:12] CC: [16:33:12] * @quidditywiki [16:33:14] * @Ameisenigel (re @Winston_Sung: By the way, it should be safe to delete [[Z1229]] for now? [16:33:15] [16:33:17] https://www.wikifunctions.org/wiki/Wikifunctions:Requests_for_deleti...) [16:34:04] Quiddity is on leave, you can refer to me (re @Winston_Sung: CC: [16:34:05] * @quidditywiki [16:34:06] * @Ameisenigel) [16:37:40] @Winston_Sung I'll briefly check with James if we're ok to delete and then let you know [16:53:09] That is a very constructive idea, and anyone that has the data could setup such lists on-wiki. (I don't have the data.) (re @amire80: 4. On Lexemes, make a lists of basic nouns that would be expected to be found in any language. (This can have several levels: 10...) [16:55:35] could we reuse the Swadesh lists? (I know there not optimal but it could be a start) (re @amire80: 4. On Lexemes, make a lists of basic nouns that would be expected to be found in any language. (This can have several levels: 10...) [16:56:02] I relay the idea on the Lexeme group! [16:56:09] Are there Swedish lists? I didn't know. (re @Nicolas: could we reuse the Swadesh lists? (I know there not optimal but it could be a start)) [17:10:36] The Simple English community did quite a bit of work on this, years ago. This is not specifically nouns, but it is a convenient on-wiki list by frequency in British English, linked to “simple” definitions. https://simple.wiktionary.org/wiki/Wiktionary:BNC_spoken_freq_01HWC (re @amire80: 4. On Lexemes, make a lists of basic nouns that would be expected to be found in any langu [17:10:37] age. (This can have several levels: 10...) [17:14:24] @Winston_Sung done by James right now [17:21:36] Big appreciated! (re @Sannita: @Winston_Sung done by James right now) [18:07:42] See *T358370* (re @Toby: New topic. Perhaps I've misunderstood, but it seems like the order of arguments chosen for a function specification actually mat...)