[16:06:01] Abstract Wikipedia would be a good use case for recipes. Like, for food. You don't need to communicate very complex ideas and it's something that anyone could enjoy or benefit from and almost anyone can contribute to [16:16:19] This reminds me of a question I forgot to ask: how accessible would the functions of Wikifunctions and Abstract Wikipedia be to outside websites? Could there be, for example, an API to allow external sites to use WF's functions and AWP's language generation? [16:17:03] Assuming it's the same level of service as current wikis it would be open to the outside world [16:17:29] Of course you are going to get results as quickly as Wikimedia's servers will let you [16:24:33] Yes, you should be right. I was thinking recipes, but also legal texts, and other formulaic texts. (re @harej: Abstract Wikipedia would be a good use case for recipes. Like, for food. You don't need to communicate very complex ideas and it...) [16:25:52] Yes, James Hare is right. I describe a demo here where an external website calls functions from Wikifunctions: https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-11-09 [16:27:19] In addition I hope it will be open enough that you can download the function implementation and run *that* on your own chips, so that you can run on your own speed [16:27:34] Could someone write a wikitext parser in Wikifunctions? [16:28:08] Can anyone write a wikitext parser in anything? [16:28:09] Naïve answer based on it being Turing complete: Yes (re @harej: Could someone write a wikitext parser in Wikifunctions?) [16:28:35] I know you can write a Lisp interpreter in Lua because Wikinews has that [16:28:39] harej: I have this project for recipes actually, and that's what I want to do by helping WikiFunctions ^^' [16:28:58] But yes, Jan is right. But it would require a lot of context though. [16:29:00] We need the reactions enabled so that I can do a laugh emoji on this (re @vrandecic: Can anyone write a wikitext parser in anything?) [16:29:25] For templates, magic words, etc [16:29:36] Legal texts could be particularly interesting since it opens up the possibility of analyzing these texts in a much more formal way than courts usually do (re @vrandecic: Yes, you should be right. I was thinking recipes, but also legal texts, and other formulaic texts.) [16:30:11] I need to write down everything I have had in mind with this project in English (most of my writings are in French) [16:30:28] Yes, I've been involved with some research on that back in the 2000s. Legal text is potentially very interesting. [16:30:52] The other thing that makes me hesitant if curious about legal texts is that, at some point, we are making interpretations of what the law says, in the process of converting from legal text to formalized representation. And most of us aren't lawyers. I don't know how much it matters in practice. [16:31:27] That's true, grammar itself can be ambiguous in some cases (re @harej: The other thing that makes me hesitant if curious about legal texts is that, at some point, we are making interpretations of wha...) [16:32:15] I would love to see actual lawyers use Wikifunctions for their own formal interpretations of the law [16:32:37] Perhaps! Or just some grammar or logics tree [16:32:52] Perhaps! Or just some grammar or logics tree sketched on paper [16:33:45] I tried doing this with a couple paragraphs of the UDHR a while ago just to see if it was doable, and it looks like it is [16:34:03] Oh no, that would need to be actual lawyers, yes [16:34:14] Agreed [16:34:21] And, forget statute for a moment. I would really, really like to see abstract representations of *regulations*, since they are more technical in nature [16:34:22] This could perhaps be inspired by, or a collaboration with even, Q64777688 [16:41:25] I was more thinking about statutes and court cases, since it would help clarify what is actual disambiguation and what is jurisprudencial creations in the interpretation of law (and make the whole process more rigorous than with natural language) (re @harej: And, forget statute for a moment. I would really, really like to see abstract representations of *regulations*, since they are m...) [16:42:46] I don't know about that. Our research was in a continental system and mostly about formalizing laws (such as tax laws) and contracts. There has been tons of work in that area since. [16:43:09] a good start would be handling unit conversions better than existing recipe sites that suggest adding 4.84774 ml of things or other ridiculous amounts (re @harej: Abstract Wikipedia would be a good use case for recipes. Like, for food. You don't need to communicate very complex ideas and it...) [16:43:46] metric recipes and imperial recipes [16:44:33] Also informal units like tablespoon : https://en.wikipedia.org/wiki/Tablespoon [16:45:22] yeah, those are pretty much always American recipe sites, I make sure to include site:.uk if I want a recipe that makes any sense to me [16:46:19] ingredients are a bit of a problem though, e.g. self-raising flour seems to be a British thing, German recipes love vanilla sugar which I hadn't come across before in the UK [16:46:49] in the United States, sugar and vanilla extract are separate products, I've never seen them mixed together either [16:47:44] That's why, to my point of view, if such a thing has to be implemented, it should not be normative recipes, but it should allow anyone to create their own recipes [16:47:45] Q3492989 is also very common in Sweden [16:47:50] I haven't heard of self-raising flour. You would add a separate agent for dough-raising [16:48:16] If I want to create my own reciipe of Hummus, I should be able to do [16:49:10] yeah, same in Germany... I also did that in the UK but I'm weird, everyone else would just buy self l self-raising flour 😆 [16:52:03] Also for the kitchen utensils, in some countries you cannot find some of them, it would be neat to help adapting in such a case to follow the recipe anyways [16:53:13] In Swedish recipes, there's a standardized conversion for the utensils to metrice unite (like teaspoon = 5 ml, table spoon = 15 ml and so on) (re @wmtelegram_bot: Also for the kitchen utensils, in some countries you cannot find some of them, it would be neat to help adapting in su...) [16:54:31] yeah, of course, I just mean that it affects the usefulness of being able to translate it if you need to adapt it to every country anyway (re @wmtelegram_bot: That's why, to my point of view, if such a thing has to be implemented, it should not be normative recipes, but it sho...) [16:57:05] The question is whether, because we have to adapt the utensils, we should create another recipe or just change the rendered text based on the user preferences / profile. [16:57:49] (with transparency, I mean, we should let know the user that the recipe normally requires utensils they don't have) [16:58:37] I'd probably should write down all my ideas and questions somewhere, and share it here [17:00:38] Also, the thing I'd love to is to add on it semantic on the rendered text, so it can be analyzed by machines [17:05:04] which utensils are you thinking of, anyway? [17:07:34] I have an example in mind, but I need to check before saying something fool ^^' [17:16:42] T323335 (http://phabricator.wikimedia.org/) – Create mw.uten convenience functions [17:26:56] Nikki: I talk about utensils specific to regions, which can be more or less hard to get depending on where you live. For example, in France, you can easily find the device for making raclettes. I bet that's not the case in other regions. [17:28:35] And of course, you can make raclette by another way than by using this device [17:29:29] Obviously, lots of utensils are hard to find in France too [17:42:28] Another example: Oroshigane (), which comes from Japan. I checked, I could not manage to find a seller in France, even via Amazon [18:28:20] I do have some project about this, with the pre-unitarian measurements in Italy (re @Nikki: a good start would be handling unit conversions better than existing recipe sites that suggest adding 4.84774 ml of things or ot...) [22:57:26] Nice video! I think I managed to do my first function as well. It is perhaps redundant, but that might be for later to figure out. Now at least I think I have a mental model of the process. https://wikifunctions.beta.wmflabs.org/wiki/Z10297 (re @wmtelegram_bot: Newsletter 93: Checking lexical forms https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-11-09 - and be...) [23:15:05] This was my first function https://wikifunctions.beta.wmflabs.org/wiki/Z10280 [23:19:57] I added a test :) (re @harej: This was my first function https://wikifunctions.beta.wmflabs.org/wiki/Z10280) [23:20:27] I encourage you to follow Mr... if you wish to live up to standard. [23:20:28] he have helped lots of people in my area with his trading technics. I'm happy that I got connected with him... click on the link below to  start earning on daily basis... (https://t.me/+3ZyT1-UaY8RmMGM0) [23:23:18] Let us know about your experience! [23:23:37] your incomplete user interface is incomplete [23:24:19] anyways my main gripe is that the button to add a test was such a small link that i thought i had to use the fallback interface even though that didn't work. i did eventually find it, the link was just small. i'm reluctant to complain about this but you did ask [23:26:10] It's confusing to have to add the function again on the test when I just clicked the link from the function. it seems like it's either redundant information or could be filled in by default. [23:28:01] Thanks for the feedback! [23:28:32] It's also very satisfying to see the test pass! :) [23:29:17] I got confused about which create link to click on the main page 😅 [23:29:28] I very much agree [23:33:35] it'd make more sense to me for the create new implementation/test buttons to be below the relevant sections (or at least connected to them in some way) [23:34:20] and I'm not sure why there's word-break: break-all in the css but it's causing things like this : https://tools-static.wmflabs.org/bridgebot/29311a7c/file_42933.jpg [23:34:35] For my part, I am really excited to try more, but I also feel like I have arrived at a restaurant while they are still installing the kitchen [23:36:40] Most certainly. It's all still on paper plates, and we only have a few things being served, and not all of the time, and the safety exits are not working yet... [23:36:49] What sort of data stores will Wikifunctions be able to pull from? [23:37:27] More specifically. Could a function have its own data store(s) that it reads from/writes to? [23:47:32] It'll be able to read from data like Wikidata, but we have no plans of having it write. There might be workflows that use the results to write, but that wouldn't happen through function calls in Wikifunctions.