[01:05:52] Toby Thanks for fixing Z20457. There’s some bug in there somewhere but I haven’t identified it. I think it may have something to do with using the radio buttons rather than the dropdown, but it may be that you have to save it in its dodgy state. I’ll look at it again in the (our) morning 😎 [01:09:08] I'm pretty sure it's a caching issue (re @Al: Toby Thanks for fixing Z20457. There’s some bug in there somewhere but I haven’t identified it. I think it may have something to...) [01:09:58] Z844 should not care about the difference between a reference in a literal bool and a reference [01:10:26] ? I'm not so sure. I removed an extra Boolean() wrap. It was comparing to something like Boolean(Boolean(True)) (re @Feeglgeef: I'm pretty sure it's a caching issue) [01:11:41] You're right. (re @Toby: ? I'm not so sure. I removed an extra Boolean() wrap. It was comparing to something like Boolean(Boolean(True))) [01:11:42] I [01:11:43] I'm not sure how that could have happened (re @Feeglgeef: I) [01:13:49] The only things I can think of are Edit source (which Al does not have installed) and a UI glitch that I can't reproduce. [01:28:54] I often get UI glitches when I try to enter things too fast for the interface to catch up. A common outcome is to leave a field empty in the save. Then when I come back to edit, I sometimes have to switch "Type" to something else so that when I switch back to the one I want, it refreshes the field entry that I need to type in. I don't think it's happened to me on [01:28:54] Booleans before, [01:28:55] but maybe I don't use them very often. [01:30:34] I've only experienced the empty field thing with natural numbers, so I assume that it's the parser not being able to complete before I save (re @Toby: I often get UI glitches when I try to enter things too fast for the interface to catch up. A common outcome is to leave a field ...) [01:33:30] I don't think it's only due to parsers though. If you're searching for a function in a composition, you often click on the function to choose it, but it doesn't show the arguments of that function that need to be filled in. [01:37:30] I think this is because Wikifunctions has to work harder to figure out what the arguments are (re @Toby: I don't think it's only due to parsers though. If you're searching for a function in a composition, you often click on the funct...) [01:38:57] Because from what I've seen on phab they have a table for the names and descriptions, so they could easily reach that [04:45:40] Would this be before or after it is on the Dagbani Wikipedia? (re @wmtelegram_bot: Ameisenigel> We're looking at test2.wikipedia.org rather than test.wikipedia.org, as it's a bit disruptive, but yes, t...) [06:40:56] Before (re @Feeglgeef: Would this be before or after it is on the Dagbani Wikipedia?) [07:18:43] This course at the university will happen today: https://www.wikifunctions.org/wiki/Wikifunctions:Project_chat/Archive/2024/10#Python_course_at_the_university [08:05:32] I think I re-created this. I don’t think it’s a bug, necessarily. What confused me is that the nested Boolean presents as an unset radio button group when the outer Boolean is expanded in Edit mode. Using one of the radio buttons to set True does not replace the nested Boolean, so it appears broken. Z20464 (re @Toby: ? I'm not so sure. I removed an extra Boolean() wrap. [08:05:32] It wa [08:05:33] s comparing to something like Boolean(Boolean(True))) [10:15:50] The chaos is great, but order will eventually reign (re @Galder: This course at the university will happen today: https://www.wikifunctions.org/wiki/Wikifunctions:Project_chat/Archive/2024/10#P...) [10:16:06] https://tools-static.wmflabs.org/bridgebot/2999803e/file_66825.jpg [10:56:24] Looks a lot like my life (re @Galder: The chaos is great, but order will eventually reign) [11:06:55] I will try to order everything later. Some good learnings from UX today [11:09:41] Was it useful as a Python teaching/learning tool? (re @Galder: I will try to order everything later. Some good learnings from UX today) [11:18:06] It was! The professor was interested (he said "next year again"). Some students had an interesting way to build their functions. The most interesting part is that adding titles is not easy/straightforward or standarized. [11:18:21] So we have lots of Untitled things now. I will order that [11:30:17] Also, as an organizer, it is difficult to track what is connected and what not [11:34:50] There is some ongoing development work on this *T376671* (re @Galder: Also, as an organizer, it is difficult to track what is connected and what not) [11:36:32] (25Q2 is the current quarter, ending December.) [14:31:56] I feel you (re @Galder: Also, as an organizer, it is difficult to track what is connected and what not) [15:30:27] is it possible to mark a function for deletion or redirection? [15:30:52] [[WF:RFD]] (re @Galder: is it possible to mark a function for deletion or redirection?) [15:57:44] 👍👍 [16:05:54] @vrandecic [[Special:ListFunctionsByTests]] would be more convenient if there was an option to not include objects that are predefined [16:09:56] Also, some functions are being listed twice? [16:24:09] do you havea link to the listing-twice thing? [16:25:20] https://tools-static.wmflabs.org/bridgebot/6034f381/file_66836.jpg [16:25:25] We just released Z20420 as a new Type -- Gregorian Calendar Date. I would like to give a small warning with this one: this is the most complex type we created so far, and it might have still some unpolished edges for dates in the very far past or future. [16:25:35] Sorry I couldn't get a screenshot (re @Feeglgeef: ) [16:26:08] Can you get me the URL of that screen? [16:26:29] actually, nevermind, I already can see it replicated [16:26:53] do you want to file the bug? [16:27:00] Yes [16:27:40] I'll also file a ticket for the predefined objects [16:29:36] thanks! [16:31:53] T381003 [16:54:09] And T381005 [16:54:31] @vrandecic on a second thought, I don't think the date object thing was a good choice [16:54:43] I think we should turn around while we still can [16:55:49] The offset thing really worries me [17:06:29] It worries me too [17:06:40] We can't get away without the offset [17:07:09] So I propose we have an object of year, month, day (re @vrandecic: We can't get away without the offset) [17:07:47] Manual conversions to a date object won't be hard, and it will just make everything much simpler [17:08:57] I am not opposed, and it sounds simpler than the current system, but it wasn't even suggested on [[Wikifunctions:Type proposals/Gregorian calendar date]] [17:09:14] My suggestion is, let's mark it again as "do not use" [17:09:30] Yes (re @vrandecic: My suggestion is, let's mark it again as "do not use") [17:09:35] Someone make the change on the type proposal page [17:09:43] I'll put my proposal in the comments (re @vrandecic: Someone make the change on the type proposal page) [17:09:46] We gather a few more reviews and comments on the proposal [17:10:01] And then we implement the new consensus [17:14:21] I un-released it and marked the type as "do not use". There is a good chance that the converters will break in an incompatible way, so please avoid working with it. We need a better idea of how to implement the conversion. Please discuss the type and its conversion here: [[Wikifunctions:Type proposals/Gregorian calendar date]] (re @vrandecic: W̶e̶ ̶j̶u̶s̶t̶ [17:14:21] ̶r̶e̶l̶e̶a [17:14:22] ̶s̶e̶d̶ ̶Z̶2̶0̶4̶2̶0̶ ̶a̶s̶ ̶a̶ ̶n̶e̶w̶ ̶T̶y̶p̶e̶ ̶-̶-̶ ̶G̶r̶e̶g̶o̶r̶i̶a̶n̶ ̶C̶a̶l̶e̶n̶d̶a̶r̶ ̶D̶a̶t̶e̶...) [17:30:09] Is there anything stopping us from defining a custom class for it? (re @vrandecic: I un-released it and marked the type as "do not use". There is a good chance that the converters will break in an incompatible w...) [17:39:41] yes, we don't have a system supporting custom classes [17:46:37] K (re @vrandecic: yes, we don't have a system supporting custom classes) [18:06:44] Does what I suggested look good? (re @vrandecic: I un-released it and marked the type as "do not use". There is a good chance that the converters will break in an incompatible w...) [18:13:57] Can I create a testing function for it? I'd like to try a few things out. (re @vrandecic: I un-released it and marked the type as "do not use". There is a good chance that the converters will break in an incompatible w...) [18:17:04] Yes. I think a small number of testing functions are OK [18:17:14] K [18:54:09] minor comment on your comment on [[Wikifunctions:Type proposals/Gregorian calendar date]] [20:54:42] Newsletter #182: [20:54:42] * WordGraph: (almost) a million forms for describing people [20:54:43] * New special page: list of functions filtered by their tests [20:54:45] * More statements! [20:54:46] * New type: day of Roman year [20:54:48] * Recent Changes in the software [20:54:49] * Next volunteers’ corner on December 9 [20:54:51] * No update next week [20:54:52] * Function of the week: is leap year [20:54:54] https://www.wikifunctions.org/wiki/Wikifunctions:Status_updates/2024-11-27 [20:54:58] This week's update is out, and includes that we have a new special page for listing functions filtered by their test status. Should the new tool be linked from prominent community hubs like the community portal? [20:55:37] It's functioneer only, so I don't think it needs to be (re @vrandecic: This week's update is out, and includes that we have a new special page for listing functions filtered by their test status. Sho...) [20:55:47] I'll add it to my Navbar, though. [20:59:07] …but 4000 will be a leap-year unless Herschel’s proposal is accepted (and it hasn’t been… yet) (re @vrandecic: Newsletter #182: [20:59:07] * WordGraph: (almost) a million forms for describing people [20:59:09] * New special page: list of functions filtered by t...) [21:02:16] ^^ (re @Al: …but 4000 will be a leap-year unless Herschel’s proposal is accepted (and it hasn’t been… yet)) [21:03:00] and none in Breton 😭 (re @vrandecic: Newsletter #182: [21:03:01] * WordGraph: (almost) a million forms for describing people [21:03:05] * New special page: list of functions filtered by t...) [21:04:01] that deficiency aside I'm not quite as keen on encouraging the complete ingest of that data at this time without further sourcing [21:05:13] thanks, fixed! (re @Al: …but 4000 will be a leap-year unless Herschel’s proposal is accepted (and it hasn’t been… yet)) [21:05:27] I hope the people reading the email version won't be too confused 😉 [21:05:46] what do you mean with sourcing? (re @mahir256: that deficiency aside I'm not quite as keen on encouraging the complete ingest of that data at this time without further sourcin...) [21:06:27] Indeed but WordGraph could be used to source existing Lexemes (re @mahir256: that deficiency aside I'm not quite as keen on encouraging the complete ingest of that data at this time without further sourcin...) [21:07:25] Isn't Google already the secondary source? (re @mahir256: that deficiency aside I'm not quite as keen on encouraging the complete ingest of that data at this time without further sourcin...) [21:08:29] the first entry in the bn text file has a string which does not exist in Wikidata, whose topic only has a French label, and which yields zero results in a Google search (re @vrandecic: what do you mean with sourcing?) [21:09:03] as such I am generally uncomfortable with just calling "Google" a source for specific lexicographical information (re @Feeglgeef: Isn't Google already the secondary source?) [21:09:27] since it's only a stepping stone away from "making shit up" [21:11:20] If you think about it, the Google AI overviews are tertiary sources (re @mahir256: as such I am generally uncomfortable with just calling "Google" a source for specific lexicographical information) [21:13:36] /s [21:13:53] having uploaded a few entries for Croatian, I do have to say that I like checking them indivdually still [21:14:09] (that this is from Google's Zürich is a separate but lesser concern) (re @mahir256: as such I am generally uncomfortable with just calling "Google" a source for specific lexicographical information) [21:14:51] still speeds me up quite a bit compared to other approaches 🙂 [21:16:10] Theoretically, couldn't we just paste TensorFlow into an implementation and create an AI? [21:17:24] @vrandecic? (re @Feeglgeef: Theoretically, couldn't we just paste TensorFlow into an implementation and create an AI?) [21:18:54] I am pretty sure that the Tensorflow neither fits within the constraints of Wikifunctions, nor does it yield an AI [21:19:15] An AI, not an LLM (re @vrandecic: I am pretty sure that the Tensorflow neither fits within the constraints of Wikifunctions, nor does it yield an AI) [21:19:48] yeah [21:26:52] What's the limit [21:27:24] Is there a size limit? Or is it just 10000ms total/9000ms execution? [22:33:43] Toby I've added a couple Z20289s to Z19814 to represent what I think the problem is in T379873 [22:37:37] what the hell (re @Feeglgeef: Toby I've added a couple Z20289s to Z19814 to represent what I think the problem is in T379873) [22:37:56] what is the point of attaching some *checks* sixteen screenfuls of broken implementations to Z19815? [22:38:14] just to demonstrate that you can be obnoxious by exploiting an already-reported bug? [22:38:30] Their looking at the wrong solution (re @lucaswerkmeister: just to demonstrate that you can be obnoxious by exploiting an already-reported bug?) [22:38:35] They're [22:39:04] Their solution would not fix the actual problem [22:39:23] I didn't think it was that much (re @lucaswerkmeister: what is the point of attaching some *checks* sixteen screenfuls of broken implementations to Z19815?) [22:40:08] hopefully I’ll never find out what you *do* consider “that much” then [22:40:29] If you refresh there should be less [22:42:02] They're also functions, not implementations (re @lucaswerkmeister: what is the point of attaching some *checks* sixteen screenfuls of broken implementations to Z19815?) [22:42:25] they were shown under “implementations” [22:42:36] Thank you (re @lucaswerkmeister: they were shown under “implementations”) [22:42:40] That's the bug [22:42:47] I don’t think it’s my fault that you’ve given Z20289 the least helpful label and description possible [22:43:11] I'll add "Test object" to the description [22:44:44] Don't we already have "sandbox objects"? If not, may that be worth implementing more systematically? [22:45:20] https://tools-static.wmflabs.org/bridgebot/69956728/animation_gif.mp4 [22:45:36] We have 2 sandbox tests, 5 sandbox implementations, and one sandbox function (re @Jan_ainali: Don't we already have "sandbox objects"? If not, may that be worth implementing more systematically?) [22:45:50] But I don't mean sandbox, I mean test [22:46:53] Perhaps we could have a predefined object for testing?