[01:22:57] Thanks! [10:43:22] I wrote some comments in https://phabricator.wikimedia.org/T384520 ... I was trying to understand what is a "resolver", and I still don't understand it. [10:45:21] The word is used quite a lot in the code of the WikiLambda extension in identifiers and comments, so I guess that it's meaningful to the software engineers. It's used only once in the translatable messages. How are translators and end users supposed to understand it? Is the term "resolver" defined anywhere? [11:00:40] I found this… 🤷‍♂️ (re @amire80: The word is used quite a lot in the code of the WikiLambda extension in identifiers and comments, so I guess that it's meaningfu...) [11:01:14] I found this… *T323158 🤷‍♂️* (re @amire80: The word is used quite a lot in the code of the WikiLambda extension in identifiers and comments, so I guess that it's meaningfu...) [11:08:46] I mean, I kind of guess that "resolver" means "a reference to an existing object or a function call", but I'm really not sure, and even if it's true, then why write "resolver" instead of "a reference to an existing object or a function call"? Are people who aren't developers of the WikiLambda extension supposed to know what a "resolver" is? [11:27:35] I’m guessing too, but since the Phabricator ticket explicitly lists those types that are “resolver types” I’m happy to infer that a resolver is an object with one of those types. There must be something else that defines which “resolvers” are excluded from the “mode selector”, but I don’t think the term is an appropriate one for a user message. To a user, a “r message> [11:27:35] esolver” (type) is just one of the options in some list (or omitted from some list, as the case may be), an (un)available type. (re @amire80: I mean, I kind of guess that "resolver" means "a reference to an existing object or a function call", but I'm really not sure, a...) [11:38:03] 5542 [11:50:04] welcome Joe [12:33:37] I'm testing https://www.wikifunctions.org/view/en/Z22664 , which was mentioned in a recent status update [12:34:18] I'm giving noun Q27022, adjective Q39338, and language Hebrew [12:34:26] And the result is void [12:34:52] There are Hebrew lexemes with senses that link to these items, if it helps. [12:34:57] What is missing? [12:36:31] Maybe it's because there is no specific Hebrew function on https://www.wikifunctions.org/view/ru/Z21733 , but there is a "default function to use". The default function would probably give a wrong result, but not void. [13:34:06] Mind creating it in a test (re @amire80: I'm giving noun Q27022, adjective Q39338, and language Hebrew) [13:34:09] ? [13:35:10] Void could literally be anything from an error in a composed function to you being rate limited by the server [13:38:07] Do we want to use float64 or Rational number for WP-facing functions? [13:38:23] I'll create a discussion for this on the project chat [13:48:51] I've done this. It's structured more like a Wikipedia RFC, which makes less room for discussion but makes it easier to come to consensus. (re @Feeglgeef: I'll create a discussion for this on the project chat) [15:31:34] https://www.wikifunctions.org/view/en/Z23231 (re @Feeglgeef: Mind creating it in a test) [16:32:14] Looks like a chopped void (re @amire80: https://www.wikifunctions.org/view/en/Z23231) [16:33:08] So looks like the feature the default wants isn't present in any of the lexemes [16:37:36] the lexemes look good tho: https://www.wikidata.org/wiki/Lexeme:L66961 https://www.wikidata.org/wiki/Lexeme:L212278 (re @Feeglgeef: So looks like the feature the default wants isn't present the lexemes) [16:40:30] ah, I see that these two lexemes are failing on Z21760 also :/ [16:40:31] digging a bit deeper, I don't see Hebrew in Z21733 (nor Breton ;) ) [16:46:04] Yes, but as I said, there is a "default" function. Isn't that good enough? (re @Nicolas: ah, I see that these two lexemes are failing on Z21760 also :/ [16:46:05] digging a bit deeper, I don't see Hebrew in Z21733 (nor Breton ;)...) [17:01:27] on Z21733 the "default" is English Z21734, thus not really a default, and it sure can't work for Hebrew :/ (re @amire80: Yes, but as I said, there is a "default" function. Isn't that good enough?) [17:04:28] Feeglgeef I'm wonrding if a true default function here could be just "agree and concat the lexemes" [17:05:44] As I said, it will probably give wrong output, but not void :) (re @Nicolas: on Z21733 the "default" is English Z21734, thus not really a default, and it sure can't work for Hebrew :/) [17:05:58] That's what I'd expect, at least 🤷🏻‍♂️ [17:07:03] yeah, "void" is not helpful but look at Z21735 it can't give any answer... (re @amire80: As I said, it will probably give wrong output, but not void :)) [18:27:26] Sure (re @Nicolas: Feeglgeef I'm wonrding if a true default function here could be just "agree and concat the lexemes") [18:28:25] It wants a positive adjective (re @Nicolas: the lexemes look good tho: https://www.wikidata.org/wiki/Lexeme:L66961 https://www.wikidata.org/wiki/Lexeme:L212278) [18:29:22] Not sure how to make them agree if we don't know the language but feel free to give that a go (re @Nicolas: Feeglgeef I'm wonrding if a true default function here could be just "agree and concat the lexemes") [18:57:21] Assuming you have both lexemes, maybe something more abstract like a list of lexeme forms. In general, the absence of a form might fall back to the lemma. Depending on the type of phrase, the fallback in the case of one missing lexeme might be tolerable; in a noun phrase, the absence of an adjective could fall back to the noun, but that depends on the role of the [18:57:21] adjective. Wheth [18:57:22] er there’s an acceptable default for the case where the head is missing, I’m not sure, but a more general noun like “mammal” or “animal” might be inferred from the item properties. This is all just “in theory”, of course. But “graceful fallback” was in the original proposal for Abstract Wikipedia, I believe. (re @Nicolas: Feeglgeef I'm wonrding if a true defau message> [18:57:23] lt function here could be just "agree and concat the lexemes") [21:07:47] 51I4 [21:08:02] 2668