[03:49:16] @vrandecic please can you add Z14326 as the equality function for Z60? [03:50:37] I'd disagree, there are multiple interpretations of what language equality means (re @u99of9: @vrandecic please can you add Z14326 as the equality function for Z60?) [03:51:03] Like you could argue dialects are the "same language" [04:01:23] As far as I understand, the only consequence of adding it to the Type is that when we create tests on functions which return languages, it shows up as the default result validator (which you are free to change if you have another as yet unwritten equality tester in mind). This would almost always be an improvement over not having any default, so having to choose a [04:01:23] function every [04:01:24] time (especially for newbies who have no idea which equality functions are on offer). But since you disagree, I'll put my case on-wiki instead for further input. [05:32:26] [[Wikifunctions:Project_chat#equality_function_for_natural_languages]] (re @u99of9: As far as I understand, the only consequence of adding it to the Type is that when we create tests on functions which return lan...) [05:38:28] Is Z13262 a duplicate of Z11390? [07:03:50] Sorry if we've been over this before Al , but why does Z873 return an object instead of a Typed list of objects? [07:08:27] I'm trying to get this to work: Z22394, and am already sure that this works: Z22393, so am wondering whether to blame Z873 or Z14404, or maybe the way they are combined in Z19274 (re @u99of9: Sorry if we've been over this before Al , but why does Z873 return an object instead of a Typed list of objects?) [08:20:34] We agreed it should be a list and you filed T367005 šŸ˜Ž (re @u99of9: Sorry if we've been over this before Al , but why does Z873 return an object instead of a Typed list of objects?) [08:39:06] Well that's extra embarrassing for me. But how is it resolved?? (re @Al: We agreed it should be a list and you filed T367005 šŸ˜Ž) [08:42:41] Its title got changed, presumably based on a misunderstanding, and the supposed problem was fixed rather than the problem stated in the description? (re @u99of9: Well that's extra embarrassing for me. But how is it resolved??) [08:46:41] Ok thanks. I've reopened it. Hopefully that's okay. (re @Al: Its title got changed, presumably based on a misunderstanding, and the supposed problem was fixed rather than the problem stated...) [08:46:53] Itā€™s not exactly a duplicate, because it explicitly returns an ā€“er result in all casesā€¦ šŸ¤·ā€ā™‚ļø (re @wikilinksbot: Z13262 ā€“ English -er form [08:46:54] Z11390 ā€“ English verb to agent noun) [08:49:29] Yes okay, I suppose it's okay to have one do the most regular predictable transform. (re @Al: Itā€™s not exactly a duplicate, because it explicitly returns an ā€“er result in all casesā€¦ šŸ¤·ā€ā™‚ļø) [08:58:57] It was an early contribution from someone whose first function was deleted as a duplicate, if I recall correctly. I wouldnā€™t object to its proposed deletion. (re @u99of9: Yes okay, I suppose it's okay to have one do the most regular predictable transform.) [09:11:19] Itā€™s okay by meā€¦ but the function should definitely be returning a typed list when itā€™s used right now, so I canā€™t see how it might be the cause of your current difficulties. (re @u99of9: Ok thanks. I've reopened it. Hopefully that's okay.) [09:13:47] True. I still have more to check. (re @Al: Itā€™s okay by meā€¦ but the function should definitely be returning a typed list when itā€™s used right now, so I canā€™t see how it mi...) [09:40:31] The Byte type has been rewritten and should work now like new: Z80 [09:41:01] I think all functions that have been using it have been rewritten (whether correctly is up in the air, but at least the new structure is being followed) [10:08:21] Still going. This error message is very unhelpful: *ErrorsUnspecified error*Error type: Unspecified error (https://www.wikifunctions.org/view/en/Z500) [10:08:22] Expected result: [ "Z1", "Z1002" ]Actual result: { "Z1K1": "Z5", "Z5K1": "Z500", "Z5K2": { "Z1K1": { "Z1K1": "Z7", "Z7K1": "Z885", "Z885K1": "Z500" }, "Z500K1": "cannot read property 'Z1K1' of undefined" } } (re @u99of9: True. I still have more to check.) [10:15:19] I kicked it around a bit, but no joy. Iā€™m guessing that the implicit fetch may be the problem butā€¦ šŸ¤·ā€ā™‚ļø (re @u99of9: Still going. This error message is very unhelpful: ErrorsUnspecified errorError type: Unspecified error [10:15:19] Expected result: [ "Z1",...) [10:23:01] I'm stuck I think. I've narrowed the gap such that this test works Z22401, and when I manually pass its output into Z22403 it works fine, but when I algorithmically pass its output into Z22394 it fails (both the last two are tests of the same function). (re @Al: I kicked it around a bit, but no joy. Iā€™m guessing that the implicit fetch may be the problem butā€¦ šŸ¤·ā€ā™‚ļø) [10:25:38] šŸ¤” And arenā€™t you getting the texts out of the Z12, just not the languages? (re @u99of9: I'm stuck I think. I've narrowed the gap such that this test works Z22401, and when I manually pass its output into Z22403 it wo...) [10:28:17] Not at the moment, this function Z19272 is just for the list of languages. (re @Al: šŸ¤” And arenā€™t you getting the texts out of the Z12, just not the languages?) [10:32:24] Yes, but Z22402 is working, so maybe itā€™s just some idiosyncrasy of Z60 thatā€™s causing the issue? šŸ¤·ā€ā™‚ļø (re @u99of9: Not at the moment, this function Z19272 is just for the list of languages. Once I've got it going I can deal more with the texts...) [10:52:44] Newsletter #189: [10:52:45] * Restricting the World, redux [10:52:46] * Invitation to the NLG SIG meeting on February 18 [10:52:48] * Recent Changes in the software [10:52:49] * In lieu of the Function of the Week: Schrƶdinger Model sub shell configuration [10:52:51] * News in Types: Byte [10:52:52] * Fresh Functions weekly [10:52:54] https://www.wikifunctions.org/wiki/Wikifunctions:Status_updates/2025-02-13 [11:18:16] Is there a typo in this "and on the Test cases one only when there were four or more"? (re @vrandecic: Newsletter #189: [11:18:16] * Restricting the World, redux [11:18:18] * Invitation to the NLG SIG meeting on February 18 [11:18:19] * Recent Changes in the softw...) [11:21:46] " removing the functionality that currently replaces the shown version of a link to an Object with its name". Can we talk more about this??? I might be misunderstanding, but does this mean we need to pages full of ZIDs? Compositions will be unreadable? [11:22:21] Maybe punctuation? ā€œā€¦and, on the test cases one, only when there were four or more.ā€ (I canā€™t say I noticed.) (re @u99of9: Is there a typo in this "and on the Test cases one only when there were four or more"?) [11:25:14] I think itā€™s only talking about wikitext links, not references in objects šŸ¤·ā€ā™‚ļø (re @u99of9: " removing the functionality that currently replaces the shown version of a link to an Object with its name". Can we talk more a...) [11:26:04] Ohhhh.. so the "one" means "table"? Ok so maybe we can rewrite "the Implementations table and on the Test cases one" as "the Implementations and Test cases tables"? (re @Al: Maybe punctuation? ā€œā€¦and, on the test cases one, only when there were four or more.ā€ (I canā€™t say I noticed.)) [11:28:28] I hope so. I haven't seen any of the wrong language bugs in links. But I've often seen it in the site notices. (re @Al: I think itā€™s only talking about wikitext links, not references in objects šŸ¤·ā€ā™‚ļø) [11:32:50] Yes, ā€œoneā€ refers to [one of the] tables, but I think it says that what was happening differed according to the table in question. Since I didnā€™t notice at the time, I donā€™t know whether that reading is correct. (re @u99of9: Ohhhh.. so the "one" means "table"? Ok so maybe we can rewrite "the Implementations table and on the Test cases one" as "the Im...) [11:36:36] Yes now I can read it, thanks. Implementations never showed it, tests only showed it when long (re @Al: Yes, ā€œoneā€ refers to [one of the] tables, but I think it says that what was happening differed according to the table in questio...) [11:38:50] Even so, thereā€™s a fair amount of conversion to be done, presumably manual? Do we just convert to {{Z|ā€¦}}? How long have we got @vrandecic? (re @u99of9: I hope so. I haven't seen any of the wrong language bugs in links. But I've often seen it in the site notices.) [11:41:24] To me it actually makes more sense not to show it unless the list is longer than 5 (smallest limit). (re @u99of9: Yes now I can read it, thanks. Implementations never showed it, tests only showed it when long) [11:43:53] Yeah, I think standard practice is to show the control but visibly disabled. (re @u99of9: To me it actually makes more sense not to show it unless the list is longer than 5 (smallest limit).) [12:01:58] Maybe. I'd like to track down the issue wherever it is before I forget what I know. But yes, I originally got into this because I was planning helper functions for Z22386, and realistically that doesn't need languages much. (re @Al: Yes, but Z22402 is working, so maybe itā€™s just some idiosyncrasy of Z60 thatā€™s causing the issue? šŸ¤·ā€ā™‚ļø) [12:02:43] How's that for a readable name!? (re @wikilinksbot: Z22386 ā€“ [pro/noun] is/are a/an/ [,] [agent n] of [,] [obj]) [12:44:09] šŸ˜Ž Iā€™m not sure we want too many specific structures of that kind. Itā€™s basically just subject plus predicate, where subject is a noun phrase and (in English) the verb phrase expresses the predicate using the copula verb and a complement. šŸ¤” We probably do want specialised functions for copulas, distinct from the standard sentence equals noun-phrase plus verb-phrase sort [12:44:09] [12:44:10] of structure, because the subject or verb can be elided in certain contexts (and your function is limited to English, which should be obvious from its label). This looks like another topic for the project chat. (re @u99of9: How's that for a readable name!?) [12:47:12] Me too. But Iā€™ll need a piece of paper! (re @u99of9: Maybe. I'd like to track down the issue wherever it is before I forget what I know. But yes, I originally got into this because ...) [12:51:43] I have no idea what we want. But I thought test Z22387 is the kind of sentence structure we could put on our userpage once wikifunctions can call wikifunctions. (re @Al: šŸ˜Ž Iā€™m not sure we want too many specific structures of that kind. Itā€™s basically just subject plus predicate, where subject is a...) [12:58:52] It's planned for next week (re @Al: Even so, thereā€™s a fair amount of conversion to be done, presumably manual? Do we just convert to {{Z|ā€¦}}? How long have we got ...) [13:01:15] I rewrote the sentence about the tables and broke it down into three sentences for more clarity (but less flow): "We fixed a bug regarding tables on Function pages' tables. The pagination controls never showed up on the Implementations table. They also showed up on the Test cases table only when there were four or more Implementations" (re @u99of9: Ohhhh.. so the [13:01:16] "one" means "tab [13:01:16] le"? Ok so maybe we can rewrite "the Implementations table and on the Test cases one" as "the Im...) [13:01:55] Thanks. I havenā€™t figured out how to do the impact assessment yet šŸ˜Ž (re @vrandecic: It's planned for next week) [13:02:14] correct, that's only for wikitext. Changed text to clarify. Thanks for the question! (re @Al: I think itā€™s only talking about wikitext links, not references in objects šŸ¤·ā€ā™‚ļø) [13:05:19] done (I saw the concerns raised, but I agree with Toby here) (re @u99of9: @vrandecic please can you add Z14326 as the equality function for Z60?) [18:59:03] Users with permissions doing things without consensus because they agree is very anti-wikimedia, but whatever (re @vrandecic: done (I saw the concerns raised, but I agree with Toby here)) [19:13:05] There was an on-wiki request with no opposition. Since it has no negative impact and can easily be changed if there is some other consensus, it was just being sensibly bold. (re @Feeglgeef: Users with permissions doing things without consensus because they agree is very anti-wikimedia, but whatever) [19:13:40] Nobody gave me enough time to respond (re @Al: There was an on-wiki request with no opposition. Since it has no negative impact and can easily be changed if there is some othe...) [19:14:34] But you have all the time you need nowā€¦šŸ¤·ā€ā™‚ļø (re @Feeglgeef: Nobody gave me enough time to respond) [21:02:23] Am I right that this is the expected behaviour? Z22413 (re @vrandecic: I think all functions that have been using it have been rewritten (whether correctly is up in the air, but at least the new stru...) [21:05:42] I think itā€™s something to do with returning a Z11 from code. (re @u99of9: Maybe. I'd like to track down the issue wherever it is before I forget what I know. But yes, I originally got into this because ...) [21:07:31] Thanks. I didn't know I did that anywhere. Is it inside one of the built-ins? (re @Al: I think itā€™s something to do with returning a Z11 from code.) [21:10:20] I havenā€™t pinned it down but I think Z22414 is the same issue and it doesnā€™t occur in Z22410. (re @u99of9: Thanks. I didn't know I did that anywhere. Is it inside one of the built-ins?) [21:18:51] Ooh yes, those are interesting tests. (re @Al: I havenā€™t pinned it down but I think Z22414 is the same issue and it doesnā€™t occur in Z22410.) [21:26:09] Again the issue arises when you get the language involved (re @u99of9: Ooh yes, those are interesting tests.) [21:40:51] Well, wellā€¦ looks like itā€™s the other way around. Z22415 extracts the Z6004K3 and disconnecting the Z803 implementation gives success in Z22394 (after a purge edit) šŸ¤” (re @u99of9: Again the issue arises when you get the language involved) [21:51:09] So Z803 is the suspect? It looked so innocent and consistent! [21:52:13] Is there a test we can put on Z22399 to fail the value by key composition? [21:55:08] Maybe the fetched lexeme forms have an irregular structure, and python is just lucky cf value_by_key (re @u99of9: So Z803 is the suspect? It looked so innocent and consistent!) [21:55:55] I was trying to think of one. I think weā€™ll need a new (?) function that compares the contents of a wrapped list (given the limitations of the test capability). (re @u99of9: Is there a test we can put on Z22399 to fail the value by key composition?) [21:58:59] Some kind of ā€œrectificationā€ may be omitted if there is no transfer to the evaluator šŸ¤·ā€ā™‚ļø The Python implementation proves there is a bug somewhere, presumably in the orchestrator. (re @u99of9: Maybe the fetched lexeme forms have an irregular structure, and python is just lucky cf value_by_key) [21:59:36] Or rather than testing the entire output we could just look for a particular feature (e.g. does it have an American English monolingual inside it?) (re @Al: I was trying to think of one. I think weā€™ll need a new (?) function that compares the contents of a wrapped list (given the limi...) [22:01:54] Yeah, but all thatā€™s ā€œinsideā€ the Z12 is a Z11-list. Weā€™d need to dig into that list with a function that takes a Z12 as input. (re @u99of9: Or rather than testing the entire output we could just look for a particular feature (e.g. does it have an American English mono...) [22:03:06] Does_Z12_have_American? [22:05:20] Thatā€™s a bit specific šŸ˜® I was thinking of something like ā€œcontains a Z11 that containsā€ (re @u99of9: Does_Z12_have_American?) [22:06:42] Denny likes variety in the weekly list :-) (re @Al: Thatā€™s a bit specific šŸ˜® I was thinking of something like ā€œcontains a Z11-list that containsā€) [22:07:31] But yes, this would be more useful. (re @Al: Thatā€™s a bit specific šŸ˜® I was thinking of something like ā€œcontains a Z11-list that containsā€) [22:09:35] If the Z12 is just passed through warts and all, will our tester always pass if it digs using python, but always fail if it digs using value_by_key? [22:11:14] Iā€™ll try something like that, then šŸ˜Ž (re @u99of9: Denny likes variety in the weekly list :-)) [22:13:38] Who knows? Itā€™s hard to imagine how Z803 fails here šŸ¤·ā€ā™‚ļø (re @u99of9: If the Z12 is just passed through warts and all, will our tester always pass if it digs using python, but always fail if it digs...) [23:04:18] Indeed it worked for both Z22401, maybe because "superset" uses code? (re @u99of9: If the Z12 is just passed through warts and all, will our tester always pass if it digs using python, but always fail if it digs...) [23:09:57] I suppose we could try wrapping Z21613? I donā€™t really approve of creating these test-only functions, but theyā€™re not completely redundant in true compositions, I suppose. (re @u99of9: Indeed it passed for both Z22401, maybe because "superset" uses code?)