[00:34:04] I think the reason why the composition isn’t working is that the elements in the list are not lists, so the element equality function shouldn’t be list equality. I tried changing it to string equality and it returned True for a pair of strings each in a single element list. That’s not what you were looking for, I guess, but I hope this makes sense. (re @Toby: Another weird [00:34:06] one IMO is Z13053 where I can't see anything I've done wrong. When I read the summary of the algorithm without exp...) [01:01:05] Ok interesting, yes that makes sense. I've just tried again and now the elements in the list are single element lists. But it's still broken 😔 (re @Al: I think the reason why the composition isn’t working is that the elements in the list are not lists, so the element equality fun...) [01:12:02] @vrandecic This reminds me. Is this a good place to suggest a few built-in functions? Z13052 and Z13036 are both hacks but both seem very useful and might be better built in. But the latter needs another version "apply2" which can call a function with two parameters, which nobody has come up with a hack for (this one is needed for Z12753 and various list accumulation or sorting f [01:12:02] unctions based on arbitrary (function defined) metrics/comparators). I don't know about apply3/4/5 etc. [03:53:15] perhaps something like "apply function with list[] of arguments" would be a general answer? [05:40:07] https://tools-static.wmflabs.org/bridgebot/78542bd4/file_57523.jpg [05:40:08] https://tools-static.wmflabs.org/bridgebot/98449a69/file_57524.jpg [05:40:10] https://tools-static.wmflabs.org/bridgebot/274c50ad/file_57525.jpg [05:40:11] https://tools-static.wmflabs.org/bridgebot/aa35da04/file_57526.jpg [05:40:13] My very-much-not-empty composition draft just got declared "empty" so won't save 😭 : https://tools-static.wmflabs.org/bridgebot/de5dfe22/file_57527.jpg [05:40:14] https://tools-static.wmflabs.org/bridgebot/b56add25/file_57528.jpg [05:40:16] https://tools-static.wmflabs.org/bridgebot/76698d70/file_57529.jpg [07:01:22] I'm getting very stuck today. I thought I'd move to a simpler function: Z12767. And as I added a new composition Z13231 and two new tests Z13230 and Z13232, all seem to fail, even the existing implementation. I thought I was getting good at list functions... [07:21:30] It hasn't changed for me. In fact I was wondering what you meant in the newsletter when you said Z10755 displays correctly as of today. Here's what I see (I've tried hard-reloading). : https://tools-static.wmflabs.org/bridgebot/2d4c723d/file_57532.jpg [07:24:45] Good question (which is true for almost all our language functions) (re @Toby: How should we handle capitalisation in morphemes? Do we need functions to do their own tests on the input capitalisation, or sho...) [07:27:38] Thanks. Since it's so general I've added it on Project Chat: https://www.wikifunctions.org/wiki/Wikifunctions:Project_chat#Handling_capitalisation_in_language_functions (re @Nicolas: Good question (which is true for almost all our language functions)) [07:29:18] Since it's general, I wouldn't treat it in each function but I'm not sure exactly how... (re @Toby: Thanks. Since it's so general I've added it on Project Chat: https://www.wikifunctions.org/wiki/Wikifunctions:Project_chat#Handl...) [08:44:35] huh, you don't have the weird "en" blob before the title (re @Toby: It hasn't changed for me. In fact I was wondering what you meant in the newsletter when you said Z10755 displays correctly as of...) [08:49:55] I guess it thinks british english isn't english 😒 [08:50:27] also I am REALLY FED UP of seeing a "language changed for this wiki" popup every time I open a link [08:53:17] (ok, maybe it's not *literally* every single time. but it feels like it) [08:53:39] I don't have it... (re @Nikki: also I am REALLY FED UP of seeing a "language changed for this wiki" popup every time I open a link) [08:55:54] which interface language do you use normally? [08:56:29] mostly French (but often Breton too) (re @Nikki: which interface language do you use normally?) [08:57:24] and clicking the links from wikibot (which change the interface language to english) don't give you the usual message about the interface language being changed? 🤨 [09:03:13] I get it on those links, and then again when clicking a normal link (which uses the language I'm actually using), then if I click a link on a page which changed it to english, it'll change it back to english, then if I go to a normal page again it'll change it back to my interface language, and using the site is basically constantly switching between languages like that, dependin [09:03:14] g on how I got to the current page [09:32:38] are you using the old vector interface? (I'm on the new vector 2022, maybe that's why I don't have it) (re @Nikki: and clicking the links from wikibot (which change the interface language to english) don't give you the usual message about the ...) [09:32:42] yes, but that popup isn't specific to vector [09:32:44] well, I don't have it at all in Vector 2022 (for some reason) (re @Nikki: yes, but that popup isn't specific to vector) [09:32:57] I get it in vector-2022 too on something like https://www.wikifunctions.org/view/en/Z10758?useskin=vector-2022&uselang=zh (in the top left corner, with a broken arrow... and that's the *fourth* position I've seen it appear on the screen) [09:33:01] strange... (re @Nikki: I get it in vector-2022 too on something like https://www.wikifunctions.org/view/en/Z10758?useskin=vector-2022&uselang=zh (in th...) [09:33:21] there needs to be a way to disable syntax highlighting when editing [09:46:31] or more specifically: a way to disable the editor that replaces the textarea, because it doesn't work for quite a few writing systems [10:04:31] the main issue I've come across is when trying to edit lines with indic scripts, the combining behaviour of the characters somehow messes up where it thinks the cursor is [10:05:48] I've heard that it doesn't work with the uls input methods either (haven't tested it though) [10:09:21] I also noticed that the line height is too small for other scripts, when I pasted something like devanagari, the parts above the line across the top got cut off [10:14:19] the font size should be relative to be page font size, not a fixed pixel size, for accessibility (and it's already too small for my liking, so maybe it should be customisable too) [10:19:13] the editor area seems to be limited to 20 lines in height, which is tiny, and when I looked at the page in vector-2022, it's also really narrow. if people are supposed to be able to edit things in their browser, maybe it should have a way to make the editor fill the screen [10:19:49] I also noticed that when the editor starts scrolling, the top of the editor is cut off, even if I scroll all the way to the top [11:05:24] Differently from Z13179, I’ve changed Z13091 so that it always returns a lower case form. I thought I might capitalize the suffix according to the capitalization of the final character in the input string, but I’m leaving that until we see an emerging consensus in the project chat topic. (re @Toby: How should we handle capitalisation in morphemes? Do we need functions to do t [11:05:25] heir own tests on the input capitalisation, or sho...) [11:23:19] By the way, although it only has twenty-odd tests, Z13177 always (eventually) fails on all tests if I change the implementation. I presume this is some limit being exceeded, because they all pass again when I refresh… but then they all fail again after Publish and I have to refresh the function page to get them all to pass. I wouldn’t say it’s a very complex implementation, [11:23:20] but… 🤷‍♂️ [11:25:06] Sorry, I mean Z13087. (re @wikilinksbot: Z13177 – English -ed form) [11:27:10] [hmm… wikilinksbot doesn’t seem to look at edits…] [11:59:08] nope only links to page, not to single edits [11:59:09] I've traced one source of my problem to what I think is misbehaviour of an inbuilt function when operating on list objects. I wrote a test that fails: Z13234 . Can someone sense-check this for me? (re @Toby: I'm getting very stuck today. I thought I'd move to a simpler function: Z12767. And as I added a new composition Z13231 and two ...) [12:00:32] Yes, I experience this a lot when the number of implementations times tests is high. (re @Al: By the way, although it only has twenty-odd tests, Z13087 always (eventually) fails on all tests if I change the implementation....) [12:17:20] I mean, if I edit the page reference, it doesn’t see that the edited comment now refers to a different page. (re @Sannita: nope only links to page, not to single edits) [12:17:44] Ah, now I've got this to work by switching the element equality function to "string equality" instead of "list equality". This is counterintuitive to me. (re @Toby: I've traced one source of my problem to what I think is misbehaviour of an inbuilt function when operating on list objects. I wr...) [12:21:36] You mean it doesn't update the reference? (re @Al: I mean, if I edit the page reference, it doesn’t see that the edited comment now refers to a different page.) [12:21:45] Yeah, it does also that [12:21:58] It's easier to put the new reference in another message [12:29:53] But this still doesn't work for an almost identical test with code point equality. Compare Z13236 to Z13237 (re @Toby: Ah, now I've got this to work by switching the element equality function to "string equality" instead of "list equality". This i...) [12:34:49] Similarly Z13230 passes but Z13232 fails (re @Toby: But this still doesn't work for an almost identical test with code point equality. Compare Z13236 to Z13237) [12:39:03] Then (finally?) these can both be made to work by switching from element code point equality to element *string* equality, even though the elements are definitely lists of code points: Z13237 Z13232 [13:26:32] Are they, though? It looks like the first element is actually a list (of code-point elements). So I guess you would need list equality on the list (element) and code-point equality on the simple code points But then… I would guess that the list is not well-formed, because its type is code point rather than Z1 🤔 🤷‍♂️ (re @Toby: Then (finally?) these can both be made [13:26:32] to work by switching from element code point equality to element *string* equality, even ...) [14:09:28] given that this channel is bridges with irc, it's probably best to avoid editing messages anyway (re @Al: [hmm… wikilinksbot doesn’t seem to look at edits…]) [14:15:31] Okay, I’ll avoid that in future. (re @Nikki: given that this channel is bridges with irc, it's probably best to avoid editing messages anyway)