[01:31:21] I'm so glad not to have to manually reload every sentence on AW!!! (re @vrandecic: Abstract Wikipedia Newsletter 244 [01:31:21] * Milestone: 4000 functions on Wikifunctions [01:31:23] * Milestone: 1000 articles on Abstract Wikipedia [01:31:24] ...) [02:52:49] Where can we order the T-shirt design you wore in Stuttgart? (re @vrandecic: Abstract Wikipedia Newsletter 244 [02:52:50] * Milestone: 4000 functions on Wikifunctions [02:52:51] * Milestone: 1000 articles on Abstract Wikipedia [02:52:53] ...) [05:37:02] Anyone interested in planning out this transition from strings to HTML... I sketched how it might work in Z33687. I have no idea what the shape of the context data should look like. For example: (Link Configuration Type) we will want to specify if each QID should be linked, and if so on which project. (Link Label Type) we will want to specify how compact the label [05:37:02] referring to th [05:37:03] at object should be: just a pronoun, surname, common name, full name, official name, etc. Perhaps there are other types of configuration required? (re @u99of9: Unfortunately it is a bit painful. This is why I prefer to do smaller tests and get them right before rolling out too much conte...) [06:50:43] I love that shirt! It's black print on a black shirt. I ordered it from Custom Ink. (re @u99of9: Where can we order the T-shirt design you wore in Stuttgart?) [06:55:27] Seemed to have been cached at a bad moment. I busted the cache for that implementation, and it straightened itself up. (re @u99of9: Results at Z33643 are incorrect, but the same inputs are correct here) [07:45:17] Perhaps on the function’s Talk page, or a page linked from there? Conceptually, each participant (“entity” instance, “class” and the implicit P31 Relation for this function) needs to supply the attributes for its anchor in the text. We also need to handle cases where natural language fuses two participants into a single span, or nests the text of one within another. (re [07:45:17] [07:45:18] @u99of9: Anyone interested in planning out this transition from strings to HTML... I sketched how it might work in Z33687. I have no idea...) [07:52:13] So a new Type should contain the qid reference along with keys for what I've called link/label configuration types? Anything else it should have? Should we build the smallest types first as we did for numerical types? (re @Al: Perhaps on the function’s Talk page, or a page linked from there? Conceptually, each participant (“entity” instance, “class” and...) [07:56:55] I can't help think we are going to need this sooner rather than later. (re @wikilinksbot: [[Wikifunctions:Type proposals/Wikifunctions object reference]] [07:56:56] T390558#10727560 – Support a light-weight enum Type alternative...) [08:00:16] Yes 😏 I always did 🤷‍♂️ (re @u99of9: I can't help think we are going to need this sooner rather than later.) [09:30:32] Seems something that could be solved with [[Wikifunctions:Type proposals/Semantic unit]] (re @Al: Perhaps on the function’s Talk page, or a page linked from there? Conceptually, each participant (“entity” instance, “class” and...) [09:32:31] 4699 [09:47:26] Whatever downstream structures may be required or selected, the AW function call has to supply the arguments to their constructors, directly or otherwise. (re @dvd_ccc27919: Seems something that could be solved with [[Wikifunctions:Type proposals/Semantic unit]]) [09:47:56] If anyone is interested, please send me a private message [10:11:20] @Admins ^^^^ [10:11:53] @vrandecic @Sannita @lucaswerkmeister (re @u99of9: @Admins ^^^^) [10:29:29] The idea with Semantic units is that they are objects that are directly constructed on AW (directly of through appropriate constructors), and then passed to a functions that renderes the fragment. Most of @u99of9 's concerns would be solved by the fact that links would be directly provided attached to the right Semantic unit (as a "Signal") (re @Al: Whatever [10:29:29] downstream structures [10:29:30] may be required or selected, the AW function call has to supply the arguments to their constructo...) [10:35:27] That is good to hear. I guess I'd like to see a design of a Signal Type, where we can brainstorm all the different things that could be signalled. (re @dvd_ccc27919: The idea with Semantic units is that they are objects that are directly constructed on AW (directly of through appropriate const...) [10:39:48] I'm still coming to grips with these proposals. In general I'm a bit unnerved by the top-down design, because as far as I understand WF types, we need to get the types of the included keys right before we introduce the overall Type. (re @u99of9: That is good to hear. I guess I'd like to see a design of a Signal Type, where we can brainstorm all the different things [10:39:48] that c...) [10:40:00] I’ve made a start with [[Talk:Z33687#Constructors]]. I’m currently agnostic about the types constructed, other than the current HTML constraint, which I’ve extended “architecturally”. (re @u99of9: That is good to hear. I guess I'd like to see a design of a Signal Type, where we can brainstorm all the different things that c...) [10:48:45] I hope you’re not including me in that! Well, yeah, we’re shooting for an Abstract Wikipedia article. I am far from convinced that a series of unconnected function calls can achieve that, but it’s a current constraint and we can certainly make progress without worrying too much about how state is transferred from one call to the next. (re @u99of9: I'm still coming to [10:48:45] grips [10:48:47] with these proposals. In general I'm a bit unnerved by the top-down design, because as far as I unders...) [10:54:15] Hello! 👋 [10:54:15] I’m looking for a few motivated people (21+) to join a new project I’m working on. [10:54:17] It’s fully online and takes about 2–3 hours a day. Nothing complicated, just simple tasks. [10:54:18] If you’re interested, just send me a “+” to @Angelina_Braccii. [10:57:09] @Angelina_Braccii and @Bianca_Schneider9 please go away. Nobody here will want to sign up to your spam. We already have a project that requires 24 hours a day around the globe. [10:57:49] done, sorry for being late (re @u99of9: @Admins ^^^^) [10:58:20] Hoi 😊 [10:58:21] Heb je soms wat vrije tijd over? [10:58:23] Ik doe iets kleins online, gewoon rustig vanuit huis, een paar uurtjes per dag. [10:58:24] Niets ingewikkelds. [10:58:26] Misschien is het ook iets voor jou 🙂 [10:58:27] Stuur me gerust een berichtje. [10:59:06] it seems that spammers are having a great time today [11:08:31] I certainly am! Personally, I'm not worried about transferring global page variables throughout all functions. But I do want the human editor to be able to suppress a link in a sentence if they know it has been linked before. (re @Al: I hope you’re not including me in that! Well, yeah, we’re shooting for an Abstract Wikipedia article. I am far from convinced th...) [11:18:52] With Semantic unit, the only required Wikifinctions type would be Semantic unit itself (and Syntactic unit for internal processing); everything would be encoded inside Semantic units (re @u99of9: I'm still coming to grips with these proposals. In general I'm a bit unnerved by the top-down design, because as far as I unders...) [11:20:39] I never said anything about “global page variables”! As I see it, every utterance represents a transition from the prior context and defines the context for the following utterance. It might be made explicit, in the manner you suggest (which is why we don’t have to worry about it too much right now), but it’s worth bearing in mind. (re @u99of9: I certainly am! Personally [11:20:39] [11:20:41] , I'm not worried about transferring global page variables throughout all functions - I don't see tha...) [11:30:15] If the first word affects the last, it seems global to me. It's only references that I'd be worried about keeping track of. I like the way that Wikipedia groups them and dumps them at the end. (re @Al: I never said anything about “global page variables”! As I see it, every utterance represents a transition from the prior context...) [11:30:36] In the current proposal, Signals would be Semantic units too (re @u99of9: That is good to hear. I guess I'd like to see a design of a Signal Type, where we can brainstorm all the different things that c...) [11:31:47] So what would a semantic unit look like that just means "I want a link from this semantic unit to Q42"? (re @dvd_ccc27919: In the current proposal, Signals would be Semantic units too) [11:48:47] Hmm… that’s a good example to think about. [11:48:48] I’m not troubled by the terminology, but I don’t see call-to-call argument passing as a “global” article (or section) level structure that is maintained from call to call (which is also not currently an option). (re @u99of9: If the first word affects the last, it seems global to me. It's only references that I'd be worried about keeping track of. I li...) [12:34:09] How can one access the "raw" text of a composition? [12:35:46] (because the "edit source" seems more like an "edit" (visual) than an "edit source"?) [12:36:04] You can't edit the source unless you use a userscript (re @adafede: How can one access the "raw" text of a composition?) [12:36:20] [[User:Feeglgeef/wikilambda_editsource.js]] [12:36:24] Is there alreay such one? (re @Feeglgeef: You can't edit the source unless you use a userscript) [12:36:31] Oh, too fast, thanks :) [12:40:17] Wonderful, thank you! : https://tools-static.wmflabs.org/bridgebot/cbe75c07/file_79475.jpg [13:06:38] Anyone know if there’s a Beta Cluster version of AW? [13:13:25] The whole thing is in beta. I don't think there is a second one publicly available. (re @wmtelegram_bot: Anyone know if there’s a Beta Cluster version of AW?) [13:14:14] Yes, this has proven useful to me. It would be nice to make it somehow official. (re @Feeglgeef: [[User:Feeglgeef/wikilambda_editsource.js]]) [13:15:59] Are you expecting that we will have to keep track of references ourselves? Or will "the system" eventually compile them for us, as long as we tag them inline? (re @Al: Hmm… that’s a good example to think about. [13:16:00] I’m not troubled by the terminology, but I don’t see call-to-call argument passing a...) [13:28:46] If it’s not the function, it must be the software. While it’s not the software, it must be the functions. The editors may provide a list of references, but there’s *c_urrently*_ no way to link them into the hypertext unless they are arguments to the function. Hence [[Talk:Z33687#Connectors]] (re @u99of9: Are you expecting that we will have to keep track of references [13:28:46] oursel [13:28:47] ves? Or will "the system" eventually compile them for us, ...) [13:30:03] I think something like: [13:30:05] `(Q58494516, {}, [], ["Q42"])` (re @u99of9: So what would a semantic unit look like that just means "I want a link from this semantic unit to Q42"?) [13:31:26] To be inserted in the field "Scope arguments" of the Semantic unit that needs to be a link (re @dvd_ccc27919: I think something like: [13:31:27] (Q58494516, {}, [], ["Q42"])) [13:53:41] Thanks, I like concrete examples. Wouldn't it be clearer if the first element pointed to a particular Wikifunction rather than the general concept of "wikilink"? That wikifunction could know it was linking to AW or to enwiki etc. It seems strange to me to expect that Wikidata would have an item for each thing we might want to do. (re @dvd_ccc27919: I think something [13:53:41] like: [13:53:42] (Q58494516, {}, [], ["Q42"])) [13:58:28] It is just an example, in which I assumed the usual convention of wikitext was used (for example, a link to enwiki would have been `"wiki:Douglas Adams"`). Or any other format we prefer (it's not even necessary it to be a string or only one argument. If for example we created a "Wikilink destination" type, it could be used instead of a string) (re @u99of9: [13:58:28] Thanks, I like concrete [13:58:29] examples. Wouldn't it be clearer if the first element pointed to a particular Wikifunction rather than t...) [14:05:39] As for why not pointing to a specific Wikifunctions, it's because of the corner cases you mentioned earlier, like when a semantic unit ends up being only part of a word, or a disjointed set of words, or something even weirder (re @u99of9: Thanks, I like concrete examples. Wouldn't it be clearer if the first element pointed to a particular Wikifunction rather [14:05:39] than t...) [15:37:01] I.e. the gadgets screen? I don't think we have an interface administrator. If we did I think we could. (re @u99of9: Yes, this has proven useful to me. It would be nice to make it somehow official.) [16:57:45] Complex numbers are still bugged (see Z33698 for example) (re @vrandecic: Back to this. I think it is fixed. Possibly other types might have the same issue (dates, maybe?). But I need to look into anoth...) [18:15:19] PSA that you can delete what the bot sends by replying to one of its messages with /delete@wikilinksbot. Please do this if you delete your original message. [18:36:11] Thank for the tip (re @Feeglgeef: PSA that you can delete what the bot sends by replying to one of its messages with /delete@wikilinksbot. Please do this if you d...)