[01:36:05] Ah yes, it’s available now! (re @cvictorovich: Still P5186 isn't returned) [01:45:48] Which function is for if-block [01:46:03] 802? [01:46:15] Found but unsure [02:09:25] https://tools-static.wmflabs.org/bridgebot/930d6233/file_66691.jpg [02:09:39] What's wrong with this implementation [02:11:03] What's the ZID? [02:11:41] Z12491 [02:11:53] I'm massively changing it to composition [02:12:10] What you see is Python [02:12:27] condition [02:12:27] contains (claims from lexeme (Argument reference ("infinitive")), Wikidata statement (, Wikidata property reference ("P5186"), Wikidata item reference ("Q2993358"))) [02:12:48] Composition Wikidata Statements are a bit bugged. If you save, I can manually edit the JSON to what you want. (re @cvictorovich: ) [02:13:47] Does this have any problems (re @cvictorovich: condition [02:13:47] contains (claims from lexeme (Argument reference ("infinitive")), Wikidata statement (, Wikidata property reference (...) [02:14:18] The statement needs subject but otherwise good (re @cvictorovich: condition [02:14:18] contains (claims from lexeme (Argument reference ("infinitive")), Wikidata statement (, Wikidata property reference (...) [02:14:31] You should be able to set that yourself [02:15:07] What should I write in subject [02:15:20] You also don't need an if here :-) (re @cvictorovich: condition [02:15:21] contains (claims from lexeme (Argument reference ("infinitive")), Wikidata statement (, Wikidata property reference (...) [02:15:35] The lexeme reference (re @cvictorovich: What should I write in subject) [02:22:59] If (contains (claims from lexeme (Argument reference ("infinitive")), Wikidata statement (Wikidata lexeme reference (Argument reference ("infinitive")), Wikidata property reference ("P5186"), Wikidata item reference ("Q2993358"))), Boolean (true), Boolean (false)) [02:23:08] Right? [02:42:31] Yes (re @cvictorovich: If (contains (claims from lexeme (Argument reference ("infinitive")), Wikidata statement (Wikidata lexeme reference (Argument re...) [02:43:37] See its current version (re @cvictorovich: Z12491) [02:45:33] But Z12493 is still not passing [02:54:21] It's expecting a Lexeme reference, but getting a string (re @cvictorovich: But Z12493 is still not passing) [02:57:40] Oh wrong type [03:01:06] Can you see it now (re @cvictorovich: But Z12493 is still not passing) [03:05:18] It's expecting a Lexeme, and getting a Lexeme reference. You need to wrap the reference with Fetch lexeme [03:10:32] If I change the input type to lexeme… [03:10:42] It would be okay? [03:11:11] It should already be lexeme? (re @cvictorovich: If I change the input type to lexeme…) [03:12:49] See my Z12440 and Z12491 [03:13:15] You might see that many tests aren’t passing [03:13:35] claims from lexeme (Argument reference ("infinitive")) the argument reference needs to wrapped in fetch lexeme [03:14:23] May you change it to create an example for me? I would apply it to other similar functions [03:15:21] In a few mins [03:15:29] Many many thanks [03:22:35] I'm working on it (re @cvictorovich: May you change it to create an example for me? I would apply it to other similar functions) [03:26:10] I'm a little lost but I have a phab ticket to report so I will do that [03:28:14] Finished? [03:28:35] No [03:28:52] Notify me when it’s ready [03:28:57] K [03:33:06] I've created T380440. It's not very important but 🤷‍♂️ [03:35:10] Finished? [03:35:46] I don't think so [03:36:10] I'm getting Gateway timeout errors. Likely a problem on WF's side?? [03:36:34] Gateway timeout. Saw that [03:36:40] Updating test cases [03:39:12] We may ask @vrandecic (re @Feeglgeef: I'm getting Gateway timeout errors. Likely a problem on WF's side??) [03:40:14] For now I update other similar functions [03:41:05] Perhaps connected to T377338? (re @cvictorovich: Gateway timeout. Saw that) [03:47:11] How can I edit an object’s JSON [03:48:11] User:מקף/wikilambda_editsource.js [03:48:22] Is what I currently use [03:48:27] I plan to fork it soon [03:49:17] Actually I'll work on that now [03:53:48] [[User:Feeglgeef/wikilambda editsource.js]] [03:56:47] Is this user- (re @Feeglgeef: User:מקף/wikilambda_editsource.js) [03:56:59] not sure [03:57:06] “Hyphen” [03:57:15] Could be? [03:57:26] Saw this name before [04:05:28] Oh I cannot edit Z12446 [04:06:38] Changing list lexeme type from arg reference to function call isn't possible [04:10:33] Yeah (re @cvictorovich: Changing list lexeme type from arg reference to function call isn't possible) [04:10:43] I edited the JSON, it should be fixed [04:10:52] Is it a bug [04:11:03] Probably 🤷‍♂️ [04:12:12] Still I cannot connect another function [04:12:28] You can see there’s a problem [04:13:08] What function do you want [04:13:12] What ZID? [04:13:35] Do it like Z12491 [04:13:51] Can you just give me a ZID for a function you want? [04:14:05] Z6825 [04:14:09] ok [04:15:45] Done [04:16:53] Perfect now [04:17:39] I gotta sleep. Goodbye! [04:24:54] Tested Z12436 with L13684, error: invalid reference [07:15:51] Rather than many functions that ask if a verb is in the 1st,2nd etc group, wouldn't it be better to just have one function that fetches the group for the input? [07:32:19] yes, but this general function could be a composition of the other small functions (re @Jan_ainali: Rather than many functions that ask if a verb is in the 1st,2nd etc group, wouldn't it be better to just have one function that ...) [07:33:05] and that's in theory because reality is more complex, for instance "ressortir" is two different verbs, one in 2nd group the other in 3rd group [07:34:05] I also get the same error (re @cvictorovich: Tested Z12436 with L13684, error: invalid reference) [07:34:23] And I would chain these together later (re @Nicolas: and that's in theory because reality is more complex, for instance "ressortir" is two different verbs, one in 2nd group the othe...) [07:35:38] Is it because of design (re @Nicolas: I also get the same error) [07:39:19] also, for this type of function, should we assume the person using this function is smart enough to know what they do or should do some verification (like checking if this lexeme is actually a verb and in French?) [07:39:53] Yes, I would add some sort of verifications (re @Nicolas: also, for this type of function, should we assume the person using this function is smart enough to know what they do or should ...) [07:39:54] I'm thinking of case like "aller" where people could select the wrong lexeme : L1324833 (noun) instead of L750 (verb) [07:41:21] Maybe create a new function: “is it a French lexeme / verb?” (re @cvictorovich: Yes, I would add some sort of verifications) [08:50:25] Sorry for the wall of answers, but I was behind just two or three days 🙂 It is awesome do see this chat so lively (and I am wondering if we should move more to on-wiki) [08:51:12] In principle sure, but I would assume that one should speak some Dagbani before contributing to the Dagbani Wikipedia. But I don't want to speak for the Dagbani community (re @Feeglgeef: If I were to contribute to the encyclopedia and use it constructively on real articles, would it be ok?) [08:52:03] I fixed that one (re @Toby: It tries to render natural number functions in compositions when they are not expanded. Example here Z13907) [08:53:15] Yes, that's the right bug. I added this case there as an example. Sorry @lucaswerkmeister for the confusion! (re @Al: …or T359231) [09:18:47] that's a great question I forgot the status off. I thought we had something like catch error, but I really need to figure it out to answer that. (re @Toby: Z20107 Is this attempt futile? Do errors propagate through a separate output stream, or can we scrutinize them with functions?) [09:28:33] Is this caused by backend issues (re @cvictorovich: Tested Z12436 with L13684, error: invalid reference) [09:35:52] It is caused by an empty string in the lexeme id. I’m not sure what causes that, but it is empty in the composition Z12441. (re @cvictorovich: Is this caused by backend issues) [09:42:59] Can I solve that (re @Al: It is caused by an empty string in the lexeme id. I’m not sure what causes that, but it is empty in the composition Z12441.) [09:50:28] In the composition, I see composition (Wikidata lexeme reference ("")), shouldn't be a key instead of a empty string ? (re @Al: It is caused by an empty string in the lexeme id. I’m not sure what causes that, but it is empty in the composition Z12441.) [09:53:25] Ultimately, it should come from the argument reference but the function being called requires a full lexeme, not a reference. (re @Nicolas: In the composition, I see composition (Wikidata lexeme reference ("")), shouldn't be a key instead of a empty string ?) [09:58:09] Maybe. I tried but the tests aren’t running successfully. (re @cvictorovich: Can I solve that) [10:04:14] Is there a sandbox where we could test things? several things seems strange in this composition, how to check simple things (like is this the correct datatype in input) [10:05:15] In theory there's the beta, but my guess is that it is too broken to test: https://wikifunctions.beta.wmflabs.org/wiki/Wikifunctions:Main_Page [10:06:28] But I’m unsure the reason (re @Al: Maybe. I tried but the tests aren’t running successfully.) [10:09:07] I’ve amended it but it tends to time out. (re @cvictorovich: But I’m unsure about the reason) [10:13:12] You directly edited its JSON? (re @Al: I’ve amended it but it tends to time out.) [10:15:03] That’s a fair point, but we already run out of time, without adding extra complexity. (re @Nicolas: Is there a sandbox where we could test things? several things seems strange in this composition, how to check simple things (lik...) [10:15:41] No, I just edited it in the normal way. (re @cvictorovich: You directly edited its JSON?) [10:25:54] Look at Z12442 (re @Al: No, I just edited it in the normal way.) [10:26:04] It seems to return false [10:26:36] While it’s indeed a group 1 verb [10:28:37] maybe it would make sense to include wikidata, testwikidata and testwikipedia, so that people have other places they can try it out without disrupting other communities (a non-test project would be useful because the test wikis can't load actual wikidata data, as far as I know) (re @vrandecic: In principle sure, but I would assume that one should speak some Dagbani [10:28:37] before contrib [10:28:38] uting to the Dagbani Wikipedia. But I don...) [10:31:58] Indeed. I didn’t say it was correct. I just corrected the empty reference. The “contains” function returns False by default. (re @cvictorovich: It seems to return false) [10:32:32] ah, oh, that's interesting! I wasn't aware of that wrinkle. Thanks for raising this! (an obvious place would be the Wikifunctions wiki) (re @Nikki: maybe it would make sense to include wikidata, testwikidata and testwikipedia, so that people have other places they can try it ...) [10:33:04] Is this another problem unrelated to previous ones I asked here (re @cvictorovich: It seems to return false) [10:35:07] How can I know if it's because Z12696 didn't get the value returned by Z6825 or something else [10:38:45] You could try creating a test for the contains function. (re @cvictorovich: How can I know if it's because Z12696 didn't get the value returned by Z6825 or something else) [10:43:30] Z6803 doesn’t allow its arguments to be a function call. Is that a bug or a current limitation? [10:44:17] @amire80 is there a way to mark messages as draft messages, to spare volunteers from working on messages that are still being fiddled with? agreed with your comments. Yes, "Upper limit" references to the field called "Upper limit of matching tests", and "Test connection status" to the "Test connection status" too. [10:50:28] Created Z20217 [10:53:20] Ah I found something weird [10:55:24] Z19300 returns all claims inside a lexeme... [10:56:05] Ah… I think I know what the problem might be. Wikidata statement now has a fourth key returned, but this is not defined in Z6003. @vrandecic? (re @cvictorovich: Created Z20217) [10:57:45] sorry I'm in a conference today but I'd love to dig deeper into this, it's important for me to really understand what is happening [10:57:47] maybe a specific online meeting on this topic? [11:00:06] Yes it's also important for my current work [11:00:54] BTW now I found that Toby is 99of9 [11:03:08] good call, fixed! (re @Al: Ah… I think I know what the problem might be. Wikidata statement now has a fourth key returned, but this is not defined in Z6003...) [11:03:49] @vrandecic What about failure of Z12442 [11:05:56] Ha. Is that good or bad? (re @cvictorovich: BTW now I found that Toby is 99of9) [11:06:23] I have something to ask about what you created... (re @Toby: Ha. Is that good or bad?) [11:06:31] I don't know, I would need more time to dig into that, and I don't have that time right now. (re @cvictorovich: @vrandecic What about failure of Z12442) [11:07:49] Toby does Z19300 prone to timeouts? [11:11:49] Yes, I have a new implementation functioning correctly now, thank you! (re @vrandecic: I don't know, I would need more time to dig into that, and I don't have that time right now.) [11:12:28] Quick thoughts: I am not sure about how trustworthy Z6803 is. That might be the problem here (which is probably used internally in "contains"). Why not a function that gets the value for P5186 given a Lexeme, and compare the result of that with the right class? (re @cvictorovich: @vrandecic What about failure of Z12442) [11:15:35] How? (re @vrandecic: Quick thoughts: I am not sure about how trustworthy Z6803 is. That might be the problem here (which is probably used internally ...) [11:16:05] A function reads the certain claim... [11:16:46] Z20218 implements Z12436 with all three tests passing. [11:17:24] Okay I'll copy it to all 3 functions [11:22:34] It seems reasonably fast. I think the test used to work, so the changes to the underlying lexeme return may have changed the structure. (re @cvictorovich: Toby does Z19300 prone to timeouts?) [11:25:02] No, copying it isn't a good choice [11:26:14] What if the given lexeme isn't reliable enough... [11:26:42] I don't think such functions should rely on outer lexemes [11:27:15] Take a look on Z12436 (re @Toby: It seems reasonably fast. I think the test used to work, so the changes to the underlying lexeme return may have changed the str...) [11:28:21] but it's hardcoded to the Lancer example? L28927 : https://tools-static.wmflabs.org/bridgebot/7455e92d/file_66696.jpg [11:28:54] THAT'S THE PROBLEM! (re @Toby: but it's hardcoded to the Lancer example? L28927) [11:29:44] Not elegant at all [11:29:47] Well, maybe it's just helping diagnose a problem. (re @cvictorovich: THAT'S THE PROBLEM!) [11:31:23] Is it? It will always return "true" (if I understand it), not sure what is tested here... (re @Toby: Well, maybe it's just helping diagnose a problem.) [11:32:04] Just like problem solving in programming contests: [11:32:13] Again for testing purposes, a sandbox somewhere would be useful (so you can check each subparts) [11:32:38] The question wants certain outputs, and you outputed them directly... [11:32:53] Sorry about that. Fixed now, I hope. (re @Toby: but it's hardcoded to the Lancer example? L28927) [11:33:02] Rather than creating proper logics in programs (re @cvictorovich: The question wants certain outputs, and you outputed them directly...) [11:34:24] Sorry. Not enough positive tests to catch my error. Fixed now, I hope. (re @cvictorovich: THAT'S THE PROBLEM!) [11:34:46] Did you encounter errors while saving? [11:35:49] Yes, it now seems to work. And although it's simpler the way you've done it (instead of wrapping it in an unnecessary if statement), it's not clear to me that there should be a difference in behaviour between the two implementations. (re @Al: Sorry. Not enough positive tests to catch my error. Fixed now, I hope.) [11:37:38] The original implementation doesn’t pick up the fourth key of Wikidata statement and it didn’t seem inclined to! 😏 (re @Toby: Yes, it now seems to work. And although it's simpler the way you've done it (instead of wrapping it in an unnecessary if stateme...) [11:41:38] Is that due to a bug, or an error in the implementation, or don't we know? My point is that the implementations now look very similar. So if it's due to an error in the implementation, I can't see it. So maybe there is a bug here? Or did you already know that? (re @Al: The original implementation doesn’t pick up the fourth key of Wikidata statement and it didn’t seem [11:41:38] inclined to! 😏) [11:42:39] What's the sense of statement rank [11:43:53] I'm not sure what you mean by "sense" here? Are you meaning to ask what a statement rank means? (re @cvictorovich: What's the sense of statement rank) [11:44:02] Sure [11:44:31] On Wikidata some statements are more true than others... [11:45:00] This might help: https://www.wikidata.org/wiki/Help:Ranking [11:45:34] Yes, and the rank indicates which one is the "truer" (re @Toby: On Wikidata some statements are more true than others...) [11:46:18] Now 12441 has 2 implements... [11:46:34] And I even made them identical... [11:47:31] Z20218 might be deleted for now [11:48:04] When the implementation was created, Z6003 had only three keys defined. Denny added the fourth to Z6003 and the new implementation picked that up automatically. (re @Toby: Is that due to a bug, or an error in the implementation, or don't we know? My point is that the implementations now look very si...) [11:48:05] There's the {{experimental}} template, which you can put at the top of qqq. I've added it recently, after a few developers asked something like this. But you do want feedback from translators who want to tinker with it anyway. (re @vrandecic: @amire80 is there a way to mark messages as draft messages, to spare volunteers from working on messages that are still being [11:48:05] fi...) [11:48:10] I don't think that helps. I think the difference was showing us either an underlying bug, or something we don't understand about one of our functions. I suggest reverting to the disconnected version which didn't quite work, until we can figure it out. (re @cvictorovich: And I even made them identical...) [11:48:39] Huh I didn't know your purpose... (re @Toby: I don't think that helps. I think the difference was showing us either an underlying bug, or something we don't understand about...) [11:48:53] Ohhh... okay, that makes sense. So this is not likely to be a bug repeated often. (re @Al: When the implementation was created, Z6003 had only three keys defined. Denny added the fourth to Z6003 and the new implementati...) [11:49:52] Z12494 is the only failing test now [11:50:04] That's what I was discussing with Al . But he seems to understand it better than I do, so maybe it's not necessary to keep the problematic one. (re @cvictorovich: Huh I didn't know your purpose...) [11:50:52] Always keep the same thing with older ID or? [11:51:29] I wouldn’t call it a bug, just a deployment issue… we need to do an impact assessment for Wikidata statement (as with any object whose definition changes). (re @Toby: Ohhh... okay, that makes sense. So this is not likely to be a bug repeated often.) [11:55:08] Z19321, for example (re @Al: I wouldn’t call it a bug, just a deployment issue… we need to do an impact assessment for Wikidata statement (as with any object...) [12:05:09] Yes, that's the one I went to as soon as you explained. So, because there's nothing "wrong" with the function or the test, I can't really see what I should change. Should I copy you and re-implement the same thing? (the implementation or the test?) (re @Al: Z19321, for example) [12:07:45] I think you just need to reset the Wikidata statement and redefine it with four keys. (re @Toby: Yes, that's the one I went to as soon as you explained. So, because there's nothing "wrong" with the function or the test, I can...) [12:13:14] I've tried to do that, but it appears even less happy. Z19321 (re @Al: I think you just need to reset the Wikidata statement and redefine it with four keys.) [12:15:03] The reference citation on Wikidata isn't brought across as well is it? L1122#P11230 [12:16:18] It’s passing for me 🤷‍♂️ (re @Toby: I've tried to do that, but it appears even less happy. Z19321) [12:16:26] Oh, nevermind - all working now. (re @Toby: I've tried to do that, but it appears even less happy. Z19321) [12:38:16] Can’t figure out why (re @cvictorovich: Z12494 is the only failing test now) [12:48:57] I've been using my own Wikilambda installation (re @Nicolas: Is there a sandbox where we could test things? several things seems strange in this composition, how to check simple things (lik...) [12:50:33] That's a good solution, but not accessible to most people (re @Feeglgeef: I've been using my own Wikilambda installation) [12:51:46] I could deploy it publicly. Would that be useful? (re @Nicolas: That's a good solution, but not accessible to most people) [12:52:13] Maybe (re @Feeglgeef: I could deploy it publicly. Would that be useful?) [13:00:51] Because it’s in group 2? (re @cvictorovich: Can’t figure out why) [13:03:03] I've made a few probability functions with the rational number type, but just realised that it's plausible we'll later introduce a probability type that is limited to the range [0,1]. Should I stop? What are the odds of this happening? [13:03:46] Imo we should make it compatible with the rational number type (Numerator/Denominator) so I see no reason to stop (re @Toby: I've made a few probability functions with the rational number type, but just realised that it's plausible we'll later introduce...) [13:03:56] Maybe categorize them for later? [13:03:58]