[01:53:26] Has something strange happened to Lexeme imports or Lexemes in code? Z27414 used to work but is now failing every test. [03:18:18] Ahhh, maybe my issue immediately above is due to this! @apine help, we still need labels from lexeme forms (at least). There may be other scenarios I haven't anticipated. (re @wmtelegram_bot: Hey, all! I jus wan to give a heads-up that labels in a Z12 are going to be stripped out of the backend after next week'...) [04:12:16] A simpler example demonstrating the issue is Z22526 (re @u99of9: Has something strange happened to Lexeme imports or Lexemes in code? Z27414 used to work but is now failing every test.) [06:36:04] Glad to see function usage integrated to functions page. This makes it easier to get info in 1 click. Example Z35972 [07:47:10] Z24139 still works, presumably because the Z12 is extracted from the Z6001 up front in the preferred implementation. The JavaScript implementations now fail. We shouldn’t be stripping any Z12s from any of the Wikidata objects. (re @u99of9: Ahhh, maybe my issue immediately above is due to this! @ apine Cory help, we still need labels from lexeme forms (at least). Th...) [07:50:25] It may not be completely clear cut about when to strip. For example a Gregorian date includes Gregorian months and their labels can be stripped. (re @Al: Z24139 still works, presumably because the Z12 is extracted from the Z6001 up front in the preferred implementation. The JavaScr...) [07:51:42] And WD objects can themselves include dates. (re @Al: Z24139 still works, presumably because the Z12 is extracted from the Z6001 up front in the preferred implementation. The JavaScr...) [07:55:47] Good point. The Z12s in Wikifunctions objects within Wikidata objects should be good to go 🤔 (re @u99of9: It may not be completely clear cut about when to strip. For example a Gregorian date includes Gregorian months and their labels ...) [07:56:56] Until I saw what happened I thought the plan was just to strip all Z2K3, Z2K4 and Z2K5 [08:00:37] But now I want the label, desc, alias keys of *every* type of object....? [08:05:16] All the Z2 keys apart from Z2K2 have always been ditched, except from Z828. (re @u99of9: But now I want to ditch the label, desc, alias keys of *every* type of object....?) [08:08:17] No, not all of them. See the reason for this thread. We need the Z12 representations of lexemes forms inside Wikidata Lexemes! (re @Al: Good point. The Z12s in Wikifunctions objects within Wikidata objects should be good to go 🤔) [08:08:56] Or maybe they don't count as WF objects?? [08:14:09] Should we be asking Cory to revert while we figure out how to specify? [08:19:48] That’s maybe the problem. Technically, everything is a Wikifunctions object, but we shouldn’t be stripping Z12s from objects that have been fetched from Wikidata. Any Z12s in “Wikifunctions” objects within the fetched objects occur when the reference is de-referenced, I believe. But I’d need to check an example… 🤔 (re @u99of9: Or maybe they don't count as WF objects??) [08:21:48] Yes. Or just exempt the fetched objects altogether (including Z2s via Z828). (re @u99of9: Should we be asking Cory to revert while we figure out how to specify?) [09:40:08] As of now, Abstract Wikipedia articles consist mostly of copular sentences (X is Y, X was Y, etc.), nouns, and adjectives. Most of the functions that support verbs are either English-only or deprecated. When can I expect useful functions for verbs to be available (can use syntactic table or syntactic phrase function)? [09:56:09] Hard to predict sorry. We could write specific functions for particular relations/verbs, so if there's some you need a lot, feel free to pitch them, but general relationships will probably need to utilise the new types. [10:08:28] The generic answer to the question is similar to the question "When will the [particular] template/module be available on this language Wikipedia?". Which of course is: when someone finds the time and motivation to put in the required effort. (re @Earldridge Jazzed: As of now, Abstract Wikipedia articles consist mostly of copular sentences (X is Y, X was Y, etc.), [10:08:28] nouns, and adjectives. Most ...) [10:11:00] First, apologies if I communicated this poorly! The plan was not to strip labels out of Z2s but out of every object containing a Z12. [10:12:36] That said, if this is causing huge and insurmountable problems, we can revert temporarily. Is there a Phabricator task for this? [10:12:41] The plan worked then. But it is too much. (re @wmtelegram_bot: First, apologies if I communicated this poorly! The plan was not to strip labels out of Z2s but out of every object cont...) [10:12:52] Yes, please revert for now, it's hard to specify what we want. [10:13:43] Okay. Please do make a task containing the calls that are failing so that we can figure out a better solution. If the solution ends up being something like, "Don't remove Z12s from Wikidata objects," we can work with that. [10:40:29] Yes it might help to know how you categorise the space of all objects, but it may be something like that. (re @wmtelegram_bot: Okay. Please do make a task containing the calls that are failing so that we can figure out a better solution. If the so...) [10:46:16] Here's a task. T436918 Others should feel free to add examples. (re @wmtelegram_bot: Okay. Please do make a task containing the calls that are failing so that we can figure out a better solution. If the so...) [11:48:17] Thank you for the task! Another question: how is this affecting people? Is this critical/breaking a lot of things? Minor enough that it could be addressed on our normal deployment cycle or in need of immediate remediation? [11:57:56] Yes, this is bad, especially for anything using lexemes on AW. I think 24-48 h would be fine, but a week would be excruciating (and the RfC will probably light up again with how bad our grammar is). (re @wmtelegram_bot: Thank you for the task! Another question: how is this affecting people? Is this critical/breaking a lot of things? Minor...) [12:30:15] Yes, I think AW is mostly surfing on the cached results, where lexemes are accessed. Please try to beat cache expiry! (re @wmtelegram_bot: Thank you for the task! Another question: how is this affecting people? Is this critical/breaking a lot of things? Minor...) [12:50:03] @u99of9 Small correction needs to be made in your recent edit to Z40711 since P1195 uses datatype "String" [13:03:00] Thanks. Because I needed to change them all at the same time I didn't get to test the ones that don't have tests. Feel free to fix, or I'll try tomorrow. (re @jsamwrites: @u99of9 Small correction needs to be made in your recent edit to Z40711 since P1195 uses datatype "String") [13:08:51] Let’s also think about fall-through results. What happens when the “best” lexeme form lacks a representation? [13:08:51] Silent failure (omission of “author” in second sentence but resolved in the first): [13:08:53] *Douglas Adams* (11 March 1952 – 11 May 2001) was an England author, humorist, and screenwriter.Douglas Adams is the of The Hitchhiker's Guide to the Galaxy. [13:08:54] I think an empty Z12 becomes a mul Z11 containing an empty string, as here: Z34952 🤔 (re @u99of9: Yes, this is bad, especially for anything using lexemes on AW. I think 24-48 h would be fine, but a week would be excruciating (...) [13:09:52] James has done a revert, so things should now get back to a working state. If there are ideas about how we can approach this issue with more nuance, let's comment on the Phab task so we can centralize the discussion. [13:10:27] Thank you team! 🙏 (re @wmtelegram_bot: James has done a revert, so things should now get back to a working state. If there are ideas about how we can approach ...) [13:14:53] Done. I made the modifications. Thanks for your modifications on other functions. Results look good. (re @u99of9: Thanks. Because I needed to change them all at the same time I didn't get to test the ones that don't have tests. Feel free to f...) [13:42:13] Nudged function and now all tests pass for Z34932 (re @wmtelegram_bot: James has done a revert, so things should now get back to a working state. If there are ideas about how we can approach ...) [13:43:31] Confirmed fixed (for now) (re @u99of9: A simpler example demonstrating the issue is Z22526) [13:48:50] Nudged; only failing for Himalayas now (returning a different result) (re @u99of9: Has something strange happened to Lexeme imports or Lexemes in code? Z27414 used to work but is now failing every test.) [13:49:28] Yes, that one is expected. (re @Al: Nudged; only failing for Himalayas now (returning a different result)) [17:05:58] *Abstract Wikipedia Neswletter 263* [17:05:59] * 5,000 Functions, passing Z40000 [17:06:00] * Recent Changes in the software [17:06:02] * News in Types: Grapheme [17:06:03] * Volunteers’ Corner on 6 September [17:06:05] * Fresh Functions weekly: 178 new Functions (!) [17:06:06] https://www.wikifunctions.org/wiki/Wikifunctions:Status_updates/2026-09-03 [17:06:08] (as usual, you get the early scoop, so you can correct all my mistakes 🙂 planning to send it out tomorrow via email and mastodon) [17:07:45] I'll work now on Mr Caesar's idea of representing dates [17:24:35] Z40962 [17:40:49] it has converters [17:54:14] Z40000 by @u99of9 is my favorite function. I hope that this will attract new contributors. (re @wikilinksbot: Z40000 – auto-infobox) [18:24:11] I can't use it yet for my fave article ( Q922 ) [18:27:10] [[:abstract:Q922]] (re @vrandecic: I can't use it yet for my fave article ( Q922 )) [23:02:25] I'm sure @jsamwrites will remedy that sometime soon! Should I add Z40000 to the default selector on AW? (re @vrandecic: I can't use it yet for my fave article ( Q922 )) [23:54:49] "The ZID is obviously the more meaningless of the two numbers, because we see them in the URL and on most pages on Wikifunctions" The rationale doesn't seem to match the claim. Did you mean meaningful? (re @vrandecic: Abstract Wikipedia Neswletter 263 [23:54:50] * 5,000 Functions, passing Z40000 [23:54:51] * Recent Changes in the software [23:54:53] * News in Types: Grapheme [23:54:54] ...)