[00:48:23] Agreed. I've added this in. (re @Al: I think Z23474 deserves a mention as an example of the function’s use.) [06:01:01] Can someone delete Z23629? I can't request on-wiki right now [06:10:59] done (re @Feeglgeef: Can someone delete Z23629? I can't request on-wiki right now) [06:11:07] Thanks [10:21:53] @Sannita thanks for [[Help:Using Wikifunctions]]. I wonder if a different name would be better though, as it seems to be specifically about errors. [10:50:30] Al I'm not sure quite what your intent was for these functions Z15818 Z15824, as they may have preceded some of our numerical types, but these two test results currently seem inconsistent to me: Z23639 Z23640. FWIW I was looking for a straightforward type check. [11:10:26] Ah, good point… We can do the job properly now. I’ll fix them up. We would need different functions for the equivalence cases. (re @u99of9: Al I'm not sure quite what your intent was for these functions Z15818 Z15824, as they may have preceded some of our numerical ty...) [11:29:17] Ah, Z15825 is for the equivalence case. I’ll create a new one for the Type. (re @u99of9: Al I'm not sure quite what your intent was for these functions Z15818 Z15824, as they may have preceded some of our numerical ty...) [11:33:15] If it's still meant to do equivalence, can I add some rational and floating equivalents? It looks like they'll fail. (re @Al: Ah, Z15824 is for the equivalence case. I’ll create a new one for the Type.) [11:34:49] As will January etc. Please, carry on! (re @u99of9: If it's still meant to do equivalence, can I add some rational and floating equivalents? It looks like they'll fail.) [11:39:28] IMO there's a kind of weaker equivalence where non-number types should never be considered equivalent (therefore I wouldn't even test them) (re @Al: As will January etc. Please, carry on!) [11:42:17] Fair point, but that’s an implementation concern. We don’t have a type for “numeric” so the function has to accept any type of object and not return True for non-Numerics. [11:42:42] Oh good. Z23644 is the way I would write that too. (re @u99of9: IMO there's a kind of weaker equivalence where non-number types should never be considered equivalent (therefore I wouldn't even...) [11:43:26] Yes, this is what I hope too. Maybe it can only be done in composition if the months turn into numbers. (re @Al: Fair point, but that’s an implementation concern. We don’t have a type for “numeric” so the function has to accept any type of o...) [11:50:25] Probably. We’ll have “has numeric type” and hardcode the types there, I guess. Presumably the other numeric types will have their own integer checks, but we could just have an “integer equivalence” helper. (re @u99of9: Yes, this is what I hope too. Maybe it can only be done in composition if in code the months turn into numbers.) [12:09:40] I was wondering the same, to be honest, but for now let's keep it this way — also because the error messages are hardcoded, and we might need to push a change before changing the page (re @u99of9: @Sannita thanks for [[Help:Using Wikifunctions]]. I wonder if a different name would be better though, as it seems to be specifi...) [12:13:44] Ideally similar pages could be made in advance of any hardcoding next time so we get to comment before it's too late. (re @Sannita: I was wondering the same, to be honest, but for now let's keep it this way — also because the error messages are hardcoded, and ...) [12:17:21] yeah, I'll suggest that next time [12:23:03] Now I've got Z23630 going you can tell me if it's abusing every principle of strong typing. (re @Al: Ah, good point… We can do the job properly now. I’ll fix them up. We would need different functions for the equivalence cases.) [13:24:21] (re @Sannita: Hi all! Our next Volunteers' Corner will be held on Monday, April 7, at 17:30 UTC. [13:24:21] If you have questions or ideas to discuss, o...) [14:23:31] Is the message "wikilambda-functioncall" used anywhere in the extension? I run `git grep -P "wikilambda-functioncall[^-]" -- ":(exclude)*.json"`, and I find no results. [14:24:21] (There are several "wikilambda-functioncall-*" messages, and they are used, but if I exclude them from the grep using `[^-]`, I find nothing.) [17:40:39] Something’s up! [17:40:39] Pretty instantaneous failures for Rational numbers in previously passing tests. [17:40:41] “Call tuples failed in returnOnFirstError. Error: FetchError: invalid json response body at http://localhost:6501/w/api.php?action=wikilambda_fetch&format=json&uselang=content&zids=Z19677 reason: Unexpected token 'u', \"upstream c\"... is not valid JSON.” [17:40:43] I thought it was just a problem with the read and display functions but it seems universal. [17:56:17] @staff Priority 0 or 1, all function evaluations failing. [17:59:04] Oof, reporting immediately [18:01:54] Thanks. Do you want a ticket? (re @Sannita: Oof, reporting immediately) [18:01:54] Yes, maybe it would be good for giving more info about the error (re @Al: Thanks. Do you want a ticket?) [18:06:48] *T391022* (re @Sannita: Yes, maybe it would be good for giving more info about the error) [18:08:11] Thanks, reported [18:08:21] hopefully someone will be working on this soon [18:20:02] Linked in [[Wikifunctions:Status#Is_anything_currently_very_broken?]] (re @Al: T391022) [19:39:29] It is probably a good idea at this time to avoid editing functions and implementations with passing tests, since this will purge the successful test results from the cache (and I guess we’ll need to purge the new failed results later). The same applies to a test case whether it is passing or failing, since the “genuine” failure will be obscured by the temporary one. [20:41:59] Back-end networking is now fixed. Sorry for the disruption, all. [20:41:59] Thanks, everyone. I am currently still getting errors but… (re @wmtelegram_bot: Back-end networking is now fixed. Sorry for the disruption, all.) [20:41:59] Al: Where are you getting errors? I can poke that and see if that's just caching or something else. [20:41:59] https://www.wikifunctions.org/wiki/Z13054 in edit [20:42:09] Yes, that’s from 2 hrs ago… (re @wmtelegram_bot: Al: Where are you getting errors? I can poke that and see if that's just caching or something else.) [20:42:29] Al: Hmm, yes. I get that to work in read mode but if I edit it even with random values it still fails. [20:52:36] Not to worry. There are a few caching issues to sort out but no new errors so far. (re @wmtelegram_bot: Al: Hmm, yes. I get that to work in read mode but if I edit it even with random values it still fails.) [20:53:46] Sorry about these. The caching issues are something we really should address. [20:54:29] Can there be a purge cache button? (re @wmtelegram_bot: Sorry about these. The caching issues are something we really should address.) [21:00:16] Feeglgeef: Certainly no button. We could write a script for it, but the cost of pressing that button on a hot function could be that the Wikipedias go blank, which'd not be fun. [21:00:29] The one thing I don’t think we can fix is errors in the read and display functions. The natural number from “4/1”, for example, gives “*Result [21:00:29] *Display function returned an unknown error.” (Running Z23660) (re @wmtelegram_bot: Sorry about these. The caching issues are something we really should address.) [21:10:11] I was just able to solve it (re @Al: The one thing I don’t think we can fix is errors in the read and display functions. The natural number from “4/1”, for example, ...) [21:12:57] Thanks. I saw. But we would need to do the same for all of the display and read functions, if they happened to have been used during the outage. (Or we let time take its course on this occasion.) (re @dvd_ccc27919: I was just able to solve it) [21:15:50] (And as James suggested, purging the cache for all the display and read functions could have a significant impact on integrated Wikipedias, so we really need to purge only the affected values.) [22:19:09] Blanking or purging sounds bad, but I imagine there could be a recalculation trigger that would gradually go through replacing the cache whenever resources were available. (re @wmtelegram_bot: Feeglgeef: Certainly no button. We could write a script for it, but the cost of pressing that button on a hot function...) [22:31:38] Still getting display error on Integer –1… shall I purge? (re @u99of9: Blanking or purging sounds bad, but I imagine there could be a recalculation trigger that would gradually go through replacing t...)