[05:22:27] Hello I have a question. If everything in a function (including the implementation and test cases) is connected. the test case originally fails. After that I changed the implementation so as to fix the issue (while connected). the test case wouldn't re-render with the new implementation. is that true? [05:22:28] or should i not change the implementation but instead create a new implementation at the first place? [05:27:44] was talking about Z34009 and Z34248 still shows failed [05:33:15] You are right that sometimes updates don't immediately get rerun. I suggest you try making a small edit to the overall function page (e.g. add an alias or description), this usually triggers a cache refresh. (re @sunmem: Hello I have a question. If everything in a function (including the implementation and test cases) is connected. the test case o...) [05:40:39] You need to correct the function Z26712 (re @Shi: But i think at least in german and at least this subject should be used with a definite article: Das Kolosseum ist ein… [05:40:40] Can i f...) [05:42:27] In this case it's Z30484 that needs to be corrected, actually (re @Shi: But i think at least in german and at least this subject should be used with a definite article: Das Kolosseum ist ein… [05:42:28] Can i f...) [05:57:40] For English you need to correct Z26088 and Z30397 (re @Shi: English too) [06:35:32] But how do we identify which items need a definite article and which don't? _'Frankreich ist ein…'_ and_ 'Die Schweiz ist ein…'_ (re @dvd_ccc27919: In this case it's Z30484 that needs to be corrected, actually) [06:38:01] It looks great in Swedish. (Where it shouldn't have a definite article.) : https://tools-static.wmflabs.org/bridgebot/1ce53115/file_80419.jpg [06:42:30] It's still an open problem, that needs to be solved in the function implementation (re @Shi: But how do we identify which lexemes need a definite article and which don't? 'Frankreich ist ein…' and 'Die Schweiz ist ein…') [07:13:31] I added tests to this function. Why are they marked as (Disconnected) [07:13:32] Did I do it right? (re @dvd_ccc27919: In this case it's Z30484 that needs to be corrected, actually) [07:15:30] The tests need to be connected manually. [07:15:31] Side note, could you please add labels to the tests? (re @Shi: I added tests to this function. Why are they marked as (Disconnected) [07:15:32] Did I do it right?) [07:16:24] (I'm not sure why we allow publishing something without any label at all.) (re @Jan_ainali: The tests need to be connected manually. [07:16:25] Side note, could you please add labels to the tests?) [07:17:58] I've labelled and connected. You'll be able to connect when you have the "functioneer" userright. (re @Shi: I added tests to this function. Why are they marked as (Disconnected) [07:17:59] Did I do it right?) [07:19:04] I assume the label should be identical to the expected result in this case? (re @Jan_ainali: The tests need to be connected manually. [07:19:04] Side note, could you please add labels to the tests?) [07:19:33] That's what I put in, although one had to be truncated. [07:19:42] I would also describe why in the 'short description' field [07:20:06] German is probably the best language to make a label in! [07:21:25] Yeah, why was it added in english? [07:22:09] That's my language default, so it was quickest for me. (re @Shi: Yeah, why was it added in english?) [07:22:20] Ill fix it [07:27:32] Is it alright like this? [07:27:32] https://www.wikifunctions.org/w/index.php?title=Z35347&diff=276162&oldid=276157 (re @u99of9: That's my language default, so it was quickest for me.) [07:40:14] Yes, although the system currently rewards us for having an identical label in English too (for example in recent changes it will now show up as untitled for me). [07:42:08] Ic (re @u99of9: Yes, although the system currently rewards us for having an identical label in English too (for example in recent changes it wil...) [07:42:38] If you want inspiration, the Italian implementation has a similar problem, and the function that tries to solve it is Z26548. Obviously you can't directly use it, since Italian has different rules that English/German (re @Shi: But how do we identify which lexemes need a definite article and which don't? 'Frankreich ist ein…' and 'Die Schweiz ist ein…') [07:45:11] I struggle to find the actual function body in the ui 😅 [07:45:46] For English one of the compositions will get it right once we have Z32645 working well. (re @dvd_ccc27919: If you want inspiration, the Italian implementation has a similar problem, and the function that tries to solve it is Z26548. Ob...) [11:53:06] When if i think that a phrase should use a more common synonymous word? [11:58:17] In just one specific language? (re @Shi: What if i think that a phrase should use a more common synonymous word?) [11:59:55] Gesamtheit > Totalität : https://tools-static.wmflabs.org/bridgebot/4520d354/file_80432.jpg [11:59:56] Yes [12:00:16] I guess it depends on if it should always be that word over the alternative in all uses in that function or only in that specific Abstract article. [12:01:41] I would say it should always prefer common words over uncommon one's [12:02:23] But i fear wikidata does not yet record how common a word is [13:02:21] There is P6191 (re @Shi: But i fear wikidata does not yet record how common a word is) [13:09:44] I imagined something like a word frequency https://en.wikipedia.org/wiki/Word_list?wprov=sfti1 (re @dvd_ccc27919: There is P6191) [13:50:13] But I see that the label for Q2445511 is Totalität [13:59:48] ? (re @dvd_ccc27919: But I see that the label for Q2445511 is Totalität) [14:06:28] Without an indication of the style of language, Z27327 chooses the lexeme that is the same as the item label (re @Shi: ?) [14:12:29] 🤔 and it looks like Q19283990 has no linked lexemes 🤷‍♂️ (re @dvd_ccc27919: Without an indication of the style of language, Z27327 chooses the lexeme that is the same as the item label) [14:58:09] Just a thought: what if we created an item "best for Abstract Wikidata", and connected lexeme senses to it via P1552? [15:29:42] I was thinking something similar, but I wonder if it would come back to bite us later. Often a synonym is preferred 9 times out of 10, but the 10th time may be because the other sense is more precise in a certain situation. (The same is possible also for frequency.) I think language style may be the better long-term solution and that we may need to model that one [15:29:43] with more nuance [15:29:43] . (re @dvd_ccc27919: Just a thought: what if we created an item "best for Abstract Wikidata", and connected lexeme senses to it via P1552?) [15:44:56] Yes, we would have to have some reason to prefer one lexeme over another, in a given context, and that is what I would expect us to capture as lexicographic data. In practice, that might mean a “neutral, semi-formal” register, which would generally be appropriate for language-specific Wikipedia articles too, unless they have a specialized domain (with a more “technical” o [15:44:56] [15:44:56] r “academic” register). [15:54:48] How does the team plan to handle internal wikilinks? Like, on the LGBTQ article, it contains wikilinks to Abstract Wikipedia articles. If a community were to import the article, would the links be automatically be changed to wikilinks within that wiki, or would they stay to Abstract Wikipedia? [16:02:15] I asked that here T421151#11866104 but there is currently no response. (re @Feeglgeef: How does the team plan to handle internal wikilinks? Like, on the LGBTQ article, it contains wikilinks to Abstract Wikipedia art...) [18:04:52] Why does Z35382 fail while Z35385, that should be completely equivalent, succeeds? [18:30:32] So i added a bunch of failing tests to Z30484. Would it be possible for someone to perhaps guide me through how i would make them pass? [18:30:34] I am totally at a loss with wikifunctions.