[00:05:25] "You don't have permission to connect an Implementation to its Function so it can be run." Do I need to be a part of a user group to link these? Z32839 and Z32840 [00:26:29] yes, Functioneer. You should apply now. [00:29:26] I've connected and added a test. Can you add labels to aid findability? (re @M: "You don't have permission to connect an Implementation to its Function so it can be run." Do I need to be a part of a user grou...) [00:31:15] Sure thing [00:44:55] There's a user on the AWP project chat who got stuck on setting a literal string to a function call. I've responded using an image to guide them in a way I hope is helpful, but perhaps the user interface could be more clear (I'm not sure in what way, though). [02:45:33] Done. Thanks a ton. It works in AW, wihch is great. Is there a way that I can add more testcases or is that also limited to Functioneers? (re @u99of9: I've connected and added a test. Can you add labels to aid findability?) [02:47:09] You can add more test cases but you cannot connect them (re @M: Done. Thanks a ton. It works in AW, wihch is great. Is there a way that I can add more testcases or is that also limited to Func...) [02:48:06] I (or another functioneer) would be willing to connect any you if you alert us here or onwiki [03:09:40] Perhaps we should make it clear on RFUG that applying for functioneer is expected for users with a few working functions created (or a large amount of implementations/test cases)? Unless you have close to no edits Functioneer is basically a guarantee. (re @u99of9: yes, Functioneer. You should apply now.) [07:21:26] Is there a function for: X is a Y in Z. [07:22:46] Z26570 (re @wmtelegram_bot: Is there a function for: X is a Y in Z.) [07:24:52] As the one who has been granting most of the functioneer rights, I remember only one case where I have not granted the rights because of missing experience with functions. (re @Feeglgeef: Perhaps we should make it clear on RFUG that applying for functioneer is expected for users with a few working functions created...) [08:33:49] Today's goal for me is to get Swedish working in it. :) (re @Ameisenigel: Z26570) [08:52:52] 2491 [09:46:34] I tried to copy and paste the Dutch implementation and just modify it for Swedish, but I guess I missed something quite fundamental because it is not working at all. Can someone spot my error? Z32858 (re @Jan_ainali: Today's goal for me is to get Swedish working in it. :)) [09:55:54] Ah, I think I got it, Z30349 that I used doesn't have any connected implementations. [09:57:27] That's odd though, becaues Z30363 is using the same function, and its tests pass. [09:59:08] Or wait, that may have been cached? Because now they don't pass. (re @Jan_ainali: That's odd though, becaues Z30363 is using the same function, and its tests pass.) [10:00:39] šŸ˜Ž That would do it! I was just going to say that if any of the calls in the list fail, the result will not be a string and the whole thing will failā€¦ which isnā€™t a great place to start šŸ¤” (re @Jan_ainali: Ah, I think I got it, Z30349 that I used doesn't have any connected implementations.) [10:30:31] Why is the second test on Z30355 failing? [10:34:53] Because itā€™s getting a reference to a lexeme, not a lexeme object. The test case is wrong, I think. (re @Jan_ainali: Why is the second test on Z30355 failing?) [10:36:07] Then why is the first test passing, they look similar to me? (re @Al: Because itā€™s getting a reference to a lexeme, not a lexeme object. The test case is wrong, I think.) [10:39:05] Yes, they look identical unless you expand them. The first test has a fetch hiding within the collapsed thing-that-looks-like-a-reference-but-might-be-a-function-call. (re @Jan_ainali: Then why is the first test passing, they look similar to me?) [10:39:40] Ah! Thanks! (re @Al: Yes, they look identical unless you expand them. The first test has a fetch hiding within the collapsed thing-that-looks-like-a-...) [10:51:07] Is there some trick to undo an edit? I would like to undo this diff, but when pressing edit, I get to the edit interface with the current values https://www.wikifunctions.org/wiki/Z30351?uselang=sv&diff=prev&oldid=261456 [10:55:30] I havenā€™t tried it in a while but you used to be able to edit the version you want to revert to and publish that. (re @Jan_ainali: Is there some trick to undo an edit? I would like to undo this diff, but when pressing undo, I get to the edit interface with th...) [11:01:07] This uses label which is unfortunate because the label on Q60 is not the official name. [11:01:08] So should we fix the label or the function? [11:01:10] Fixing the function to use the preferred statement of official name for the language seems like a good idea to me. (re @wikilinksbot: Z26570 ā€“ State location using entity and class) [11:02:42] Wikidata labels don't have to be the official names. I believe there is a property to give the official name. (re @Npriskorn: This uses label which is unfortunate because the label on Q60 is not the official name. [11:02:43] So should we fix the label or the functi...) [11:05:40] In this specific case the best way forward is probably to adjust the test to fit the current label [11:07:18] Although if we use the official name we would need to be able to fallback to the label since official names are often not the same as the preferred one in the rendered language. For example, for a text in Swedish, if we use official name for Q36600#P1448, we would get 's-Gravenhage instead of the expected Haag. (re @Npriskorn: This uses label which is unfortunate [11:07:18] because the labe [11:07:19] l on Q60 is not the official name. [11:07:20] So should we fix the label or the functi...) [11:09:00] fixed the label, now it fails because of a missing article : https://tools-static.wmflabs.org/bridgebot/11aeb3ee/file_78966.jpg [11:11:55] I meant that we first look at the best statement in the target language and fall back to label. [11:11:56] In the case of The Hague there is no swedish statement for official name. (re @Jan_ainali: Although if we use the official name we would need to be able to fallback to the label since official names are often not the sa...) [11:14:18] Nor should it be. p1448 says "official name of the subject in its official language(s)" so we shouldn't store any names that are not it the subjects languages there. (re @Npriskorn: I meant that we first look at the best statement in the target language and fall back to label. [11:14:19] In the case of The Hague there i...) [11:15:29] (Even in Dutch I would argue that in most cases you would expect Den Haag rather than the actual official name.) (re @Jan_ainali: Although if we use the official name we would need to be able to fallback to the label since official names are often not the sa...) [11:29:08] Yes, itā€™s a tricky oneā€¦ In general, I would expect the ā€œsubjectā€ to have the same label as its item, where that exists in the target language. Even if we are somehow falling back to a native language, Iā€™d expect to see the item label specified for that language. But if there is an official name (in that language?) that differs from that label, it is reasonable to surfac [11:29:08] [11:29:10] e that as well, if only parenthetically (with, of course, a link to the Wikidata statement that is the source of this information). [11:35:55] Is it the similar case with Z32858? [11:35:56] I still think it is almost impossible to understand the error messages under the details. Do we know if that is being improved? (re @Al: Because itā€™s getting a reference to a lexeme, not a lexeme object. The test case is wrong, I think.) [11:43:59] Thatā€™s just the same error, from cached results (I think) [11:44:01] Looks correct to me! (https://www.wikifunctions.org/view/en/Z32855?call=%7B%22Z1K1%22%3A%22Z7%22%2C%22Z7K1%22%3A%22Z32855%22%2C%22Z32855K1%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q1754%22%7D%2C%22Z32855K2%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q515%22%7D%2C%22Z32855K3%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q34%22%7D%7D) šŸŽ‰ (re [11:44:01] @Jan_ai [11:44:02] nali: Is it the similar case with Z32858? [11:44:04] I still think it is almost impossible to understand the error messages under the details. ...) [11:47:10] I remember a talk from a librarian at Q133846580 [11:47:11] and they said it is actually wrong in practice to translate things like Q2981. This is also why there is P1705 [11:47:47] Donā€™t know if the rules for monuments are different than the ones for cities [12:02:55] It's so interesting that we as a culture can go to space etc, but we can't properly source statements of what things should be called or handle that gracefully across languages. [12:03:04] It depends. Thatā€™s why we have labels in different languages. No anglophone would call Notre-Dame ā€œour Ladyā€ but very few would call the Eiffel Tower ā€œla Tour Eiffelā€. Oddly(?), its P1705 is Ā« Notre-Dame de Paris Ā» but its French label is Ā« cathĆ©drale Notre-Dame de Paris Ā», which is its frwiki article title. (re @adafede: Donā€™t know if the rules for monuments are message> [12:03:05] different than the ones for cities) [12:03:20] And now it works! : https://tools-static.wmflabs.org/bridgebot/0cfba69d/file_78967.jpg [12:03:37] Oh double city :) [12:03:38] Most people don't seem to care about the details, but you can't build a nuclear bomb without getting into details. [12:06:22] ā€¦nor a Wikipedia! šŸ˜ (re @Npriskorn: Most people don't seem to care about the details, but you can't build a nuclear bomb without getting into details.) [12:07:13] names are so so hard. Maybe we should exclude it from the basics so we can move forward without their complexity? (re @Al: It depends. Thatā€™s why we have labels in different languages. No anglophone would call Notre-Dame ā€œour Ladyā€ but very few would ...) [12:09:18] Certainly not! Thereā€™s a reason why most Wikipedia editions and articles clarify these points in their very first sentence. (re @Npriskorn: names are so so hard. Maybe we should exclude it from the basics so we can move forward without their complexity?) [12:09:38] But why? Z32855 is giving me what I excpect : https://tools-static.wmflabs.org/bridgebot/289507e9/file_78968.jpg [12:13:12] I think it was just a glitch, now it works... [12:14:08] 3149 [12:17:22] Four hours, of which perhaps 90% was frustrated head scratching trying to figure out what was wrong, to build one sentence is not really motivating, though. [12:20:42] Perhaps you should think of it as many thousands of beautiful Swedish sentences, some of which may even be lacking in svwiki? šŸ¤·ā€ā™‚ļø (re @Jan_ainali: Four hours, of which perhaps 90% was frustrated head scratching trying to figure out what was wrong, to build one sentence is no...) [12:24:09] Q928 now has a bunch of "Could not acquire WASI runner within time limit" errors, regardless of the language selected. They weren't there when I created the article. Using Special:Purge did not help. [12:26:41] Fair point. But it doesn't make it more fun while knee deep into it. Debugging has become more user friendly to motivate me to contribute more often than every three months (when I effectively forgot the pain from the last try). (re @Al: Perhaps you should think of it as many thousands of beautiful Swedish sentences, some of which may even be lacking in svwiki? [12:26:41] šŸ¤·ā€...) [12:31:45] Yesā€¦ there needs to be a proper debugging mode. More informative errors can help, but these were not available when most of the experimental functions were created. The new v2 composition language should help, but it remains to be seen how much. (re @Jan_ainali: Fair point. But it doesn't make it more fun while knee deep into it. Debugging has to become more user friendly [12:31:45] to motivate me t...) [12:32:06] I agree that the debugging info rarely explains itself clearly enough to know what to do. Sorry we didn't get to help you more quickly. Another option is to break through the barrier by persisting until you have made all possible types of mistake, so you'll recognise them again! But seriously, yes, we need Swedish! (re @Jan_ainali: Fair point. But it doesn't make it [12:32:06] more fun whi [12:32:07] le knee deep into it. Debugging has to become more user friendly to motivate me t...) [12:41:00] Well, you have to sleep some time! šŸ˜ I was tinkering around before v2 landed, but I never got round to trying out `Z854`. Has anyone? (re @u99of9: I agree that the debugging info rarely explains itself clearly enough to know what to do. Sorry we didn't get to help you more q...) [12:53:03] Generated English article for Q928 with four "Could not acquire WASI runner within time limit" errors : https://tools-static.wmflabs.org/bridgebot/556fab3e/file_78970.jpg [12:54:56] They don't happen when the functions are run by themselves in Wikifunctions [12:55:32] Then I'm interested in hearing a solution to the label/official name not being suitable for AW purposes on some items. (re @Al: Certainly not! Thereā€™s a reason why most Wikipedia editions and articles clarify these points in their very first sentence.) [12:58:19] Imagine how many hours have been wasted on nuclear bombs development and maintenance šŸ™ˆ (re @Jan_ainali: Four hours, of which perhaps 90% was frustrated head scratching trying to figure out what was wrong, to build one sentence is no...) [12:59:29] Yep. Bad user experience can kill any volunteer based project. (re @Jan_ainali: Fair point. But it doesn't make it more fun while knee deep into it. Debugging has to become more user friendly to motivate me t...) [13:03:27] Yesā€¦ it shouldnā€™t happen, but once it does it can take a while to clear. Basically (I think) the back-end server gets overloaded and just starts saying ā€œthereā€™s no point waiting until I get to you, because I wonā€™t.ā€ (re @Earldridge Jazzed: They don't happen when the functions are run by themselves in Wikifunctions) [13:08:18] Mostly repainting the deck today. šŸŽƒ now. Great to see so much action in recentchanges. (re @Al: Well, you have to sleep some time! šŸ˜ I was tinkering around before v2 landed, but I never got round to trying out Z854. Has anyo...) [13:10:29] Sorry to hear, just be patient and report all problems on wiki or on Phabricator (re @Jan_ainali: Fair point. But it doesn't make it more fun while knee deep into it. Debugging has to become more user friendly to motivate me t...) [13:10:49] We're trying our best to give you the best experience [13:10:51] Yes, the launch was premature but new contributors seem to be tolerant šŸ‘ (re @u99of9: Mostly repainting the deck today. šŸŽƒ now. Great to see so much action in recentchanges.) [13:19:35] Itā€™s appreciated! šŸ™ One problem, though, is that people who donā€™t get something to work donā€™t always know what their problem is, so they just walk away (presumably). Is there any way we might be able to analyse the failed attempts? (re @Sannita: We're trying our best to give you the best experience) [13:21:11] Engaging with them is one (expensive in terms of time) way (re @Al: Itā€™s appreciated! šŸ™ One problem, though, is that people who donā€™t get something to work donā€™t always know what their problem is,...) [13:21:23] Maybe a survey? [13:41:19] I supposeā€¦ but they are invisible to the community unless they make an edit somewhere. I donā€™t think weā€™re terrible at helping people who complain or ask for help. I was thinking more of a log of function calls that fail during an edit session that is not published. If we knew which functions people were trying to use, we could focus on improving those. Thereā€™s not a lot [13:41:19] [13:41:20] we can do if their language lacks adequate support, but maybe the default functions could generate a link to a page encouraging them to leave a comment? Or any page with an error could have such a link? This, for example, is not helpful: [13:41:22] ā€œWikifunctions returned a failed response: Could not acquire WASI runner within time limitā€ [13:41:23] ā€¦but it shouldnā€™t be hard to add a reporting link to the message (with a better explanation in the linked page). [13:41:25] I think we already have a ticket for more error detail, so Iā€™ll link to this comment there (when I find it, which is another challenge). (re @Sannita: Engaging with them is one (expensive in terms of time) way) [13:46:48] T420747 (re @Al: I supposeā€¦ but they are invisible to the community unless they make an edit somewhere. I donā€™t think weā€™re terrible at helping p...) [14:16:47] I found Z32199 and Z32204 while adding new functions to the catalogue, but they behave strangely. It outputs a monolingual text with an argument reference (specifically Z26107K2). [14:41:47] I think my problem mainly was that I didn't know what the problem was šŸ˜… [14:41:47] I'll write something onwiki next time I am confused. (re @Sannita: Sorry to hear, just be patient and report all problems on wiki or on Phabricator) [14:42:47] Thanks. Iā€™ll look into those later if they donā€™t get fixed before I get to them! (re @Earldridge Jazzed: I found Z32199 and Z32204 while adding new functions to the catalogue, but they behave strangely. It outputs a monolingual text ...) [14:43:15] T420747#11760816 (re @Earldridge Jazzed: T420747) [14:49:40] It's actually a bug in Z27080 when Z21583 is used anywhere in the "String" argument. In other implementations of Z26107, I get this error message: [14:49:40] ``` [14:49:41] Error in evaluation (function call: "'ZReference' object has no attribute 'Z60K1'")``` [14:49:43] (re @Al: Thanks. Iā€™ll look into those later if they donā€™t get fixed before I get to them!) [14:55:19] Yeah, I think itā€™s a bug in v2 but Iā€™ll have to look into it to be sure. (re @Earldridge Jazzed: It's actually a bug in Z27080 when Z21583 is used anywhere in the "String" argument. In other implementations of Z26107, I get t...) [15:13:54] I'll actually suggest that and see if it's feasible (re @Al: I supposeā€¦ but they are invisible to the community unless they make an edit somewhere. I donā€™t think weā€™re terrible at helping p...) [15:15:45] Thanks. I mentioned you when I put it in T420747#11760816, so you should be subscribed. (re @Sannita: I'll actually suggest that and see if it's feasible) [15:35:39] Iā€™ve made a quick fix to Z21583 so that it always returns a simple string, rather than also providing the language tag for the language it falls back to. Because of a recent v2 change or bug, the language for the Monolingual text is being identified by reference, which is not comparable with the language tag provided in the target language argument. All cases were going [15:35:39] down th [15:35:40] e exception route and this failed because a reference has no Z60K1. (re @Earldridge Jazzed: It's actually a bug in Z27080 when Z21583 is used anywhere in the "String" argument. In other implementations of Z26107, I get t...) [15:46:33] Yep, I am, thanks (re @Al: Thanks. I mentioned you when I put it in T420747#11760816, so you should be subscribed.) [16:19:31] I've added this to [[WF:Functioneers]]: [16:19:32] "Users are generally expected to apply for functioneer rights if their work on the wiki leads to frequent dependence on functioneers. Contributions to Abstract Wikipedia and the ability to work with multilingual functions may be seen as contributing factors towards granting the rights." (re @Ameisenigel: As the one who has been granting most of the functioneer [16:19:32] rights, I remember [16:19:34] only one case where I have not granted the rights be...) [16:48:29] A more robust fix is Z32874 (re @Al: Iā€™ve made a quick fix to Z21583 so that it always returns a simple string, rather than also providing the language tag for the l...) [17:05:40] Looks like the folks on enwiki are very mad about this (and very very doubtful about the project in general) (re @Feeglgeef: You also stole the work of contributors to enwiki by not attributing the content as the license requires when you didn't transfo...) [17:14:31] Quick question: When editing an implementation, does the "Try this implementation" run what I just changed (but not saved yet) or does it run whatever was there before I pressed edit? : https://tools-static.wmflabs.org/bridgebot/27a36c73/file_78972.jpg [17:17:14] It should run the changed version but, you know, no guarantees! (re @Jan_ainali: Quick question: When editing an implementation, does the "Try this implementation" run what I just changed (but not saved yet) o...) [17:18:12] Why no guarantees? It is known to be inconsistent? (re @Al: It should run the changed version but, you know, no guarantees!) [17:20:30] There was a bug some time ago and I donā€™t know whether it has been fixed. If it doesnā€™t take account of your latest changes, it is certainly a bug. [17:24:30] This always seemed to be specific to test cases, but it was what I was worrying about: *T406784* (re @Jan_ainali: Why no guarantees? It is known to be inconsistent?) [17:25:51] Oh, that's even worse! (re @Al: This always seemed to be specific to test cases, but it was what I was worrying about: T406784)