[02:27:49] At what stage would you expect it simplified? I'd actually hope it wasn't unless it went into code. (re @Al: I noticed when setting up Z29297 that a Rational number in a typed pair appears not to be simplified. Is that a known bug, or do...) [05:40:03] At Z24302 there seems to be more in the "actual result" than there used to be. To me it seems harder to read, although maybe I'm just not used to it yet. : https://tools-static.wmflabs.org/bridgebot/e9c82935/file_75158.jpg [05:43:00] Similarly at Z24177 when you expand the first dropdown of a result, it seems to have a whole bunch of details that maybe weren't visible before? : https://tools-static.wmflabs.org/bridgebot/8b9b36a4/file_75159.jpg [08:40:21] I was talking about when it goes into code. Object equivalence is only implemented in Python, but it doesn’t see 1/6 (in a pair) as equal to 2/12 (paired with the same object), because it remains an object rather than being converted to a fraction. (re @u99of9: At what stage would you expect it simplified? I'd actually hope it wasn't unless it went into code.) [09:07:46] It’s the de-referenced type, Z1.Z2K2, minus all languages but English. I don’t know why it’s not being replaced by a reference to Z1, but it’s happening by default in Z810. (re @u99of9: At Z24302 there seems to be more in the "actual result" than there used to be. To me it seems harder to read, although maybe I'm...) [09:10:50] Right, I see what you mean, and have reproduced it. I think I'd be happy with simplification here. But I haven't used pairs much (at all?), so I'm not really sure what we will usually want. (re @Al: I was talking about when it goes into code. Object equivalence is only implemented in Python, but it doesn’t see 1/6 (in a pair)...) [09:15:37] So is this a mistake? Interestingly the actual result at Z12719 includes lots of non-english labels : https://tools-static.wmflabs.org/bridgebot/aab5c289/file_75162.jpg [09:26:44] It feels like a mistake, yes. I just can’t be sure it’s new. Z18475 doesn’t do it, so we could wrap with that, I suppose. (re @u99of9: So is this a mistake? Interestingly the actual result at Z12719 includes lots of non-english labels) [10:00:54] Yeah… I think it may be just a Z1-context thing, like untyped lists 🤔 (re @u99of9: Right, I see what you mean, and have reproduced it. I think I'd be happy with simplification here. But I haven't used pairs much...) [10:12:02] Abstract Wikipedia Newsletter 225 [10:12:03] * First round of voting for naming the wiki for abstract content closed [10:12:04] * Embedded Wikifunctions on Bengali Wikipedia and seven more Wiktionaries [10:12:06] * Wikifunctions in Action on Wiktionaries [10:12:07] * Recent Changes in the software [10:12:09] * Recording of November Volunteers’ Corner [10:12:10] * Recording of the Mentor Me! Session with Tyap Wikimedians [10:12:12] * Fresh Functions weekly: 21 new Functions [10:12:13] https://www.wikifunctions.org/wiki/Wikifunctions:Status_updates/2025-11-05 [10:33:48] I added a Swedish test for Z27885. The connected implementation fails, somehow it gets L579263-F3 instead of what I would hope: L579263-F1. [11:51:36] Is there a single place to see all projects that has function calls available? (re @vrandecic: Abstract Wikipedia Newsletter 225 [11:51:36] * First round of voting for naming the wiki for abstract content closed [11:51:37] * Embedded Wikifuncti...) [12:47:13] https://noc.wikimedia.org/conf/highlight.php?file=dblists/wikifunctionsclient.dblist (re @Jan_ainali: Is there a single place to see all projects that have function calls available?) [13:12:30] Hmm… selecting the “best” form from a lexeme is just a hard problem. In practice, I would have thought it varies according to the language and lexical category, at least. If common gender and/or indefinite is specified without plural, Z27410 returns “genitiv”, otherwise, “genitiva” is preferred (for whatever reason). For headings and titles more generally, I do wond [13:12:30] [13:12:31] er whether it makes more sense to prefer the item label, and resort to constructing a label from the lexeme only when the concept has no established “title” in the required language. This would be more consistent with the case where the concept actively links to an article. (re @Jan_ainali: I added a Swedish test for Z27885. The connected implementation fails, somehow it [13:12:32] gets [13:12:33] L579263-F3 instead of what I would hope: ...) [14:54:14] Now you can add a Swedish implementation in Z29317 (re @Jan_ainali: I added a Swedish test for Z27885. The connected implementation fails, somehow it gets L579263-F3 instead of what I would hope: ...) [14:55:51] Z27890 is an example on why taking the Wikidata label isn't always the best idea (re @Al: Hmm… selecting the “best” form from a lexeme is just a hard problem. In practice, I would have thought it varies according to th...) [15:08:09] "genitiva" is preferred because has less grammatical features (it's a naive heuristic to sort out forms with unwanted features, like "contraction" or "obsolete") (re @Al: Hmm… selecting the “best” form from a lexeme is just a hard problem. In practice, I would have thought it varies according to th...) [17:16:32] Yes… I’m not sure there is a “best idea”, really. At the very least, we should be able to supply the required text in the function call, for cases where the “best guess” happens to be wrong, or not quite right. (re @dvd_ccc27919: Z27890 is an example on why taking the Wikidata label isn't always the best idea) [17:18:40] Thanks for clarifying. I looked at the code and decided not to try and work out the precise reason 😎 (re @dvd_ccc27919: "genitiva" is preferred because has less grammatical features (it's a naive heuristic to sort out forms with unwanted features, ...)