[02:46:21] Yes. ComplSlashPartLast will handle English phrasal verb particles moving around in front of and after the object? Good to know. [07:29:44] yep [07:33:27] * inariksit 's article on testing GF was accepted to NLCS \o/ [10:01:23] also fixed a bug in the test case generation, which means the sentences are still nonsensical as hell, but now there's fewer of them [10:25:26] I wonder if one can go insane from reading too much "it is nothing you without who won't be without yourself a worm without something" sentences [11:30:34] *** Joins: chat7551 (~user715@2405:204:330c:b01f:6dfb:582a:5f5f:d758) [11:31:24] *** Parts: chat7551 (~user715@2405:204:330c:b01f:6dfb:582a:5f5f:d758) () [11:48:08] hah, found new bug in english [11:48:11] TestLang: UseCl (TTAnt TPres ASimul) PPos (GenericCl (ComplSlash (SlashVV can8know_VV (Slash3V3 add_V3 (UsePron it_Pron))) (UsePron i_Pron))) [11:48:12] TestLangEng: one can me add to it [11:48:22] Lang: UseCl (TTAnt TPres ASimul) PPos (GenericCl (ComplSlash (SlashVV must_VV (Slash3V3 give_V3 (UsePron it_Pron))) (UsePron i_Pron))) [11:48:22] LangEng: one must me give it [11:48:33] and more severely, [11:48:34] Lang: UseCl (TTAnt TPres ASimul) PPos (GenericCl (ComplSlash (SlashVV want_VV (Slash3V3 give_V3 (UsePron it_Pron))) (UsePron i_Pron))) [11:48:34] LangEng: one wants me to give it [11:48:46] that's not the same as "one wants to give me it" [15:09:57] *** Joins: ConcreteBot (~ConcreteB@179.210.42.93) [15:09:57] [telegram] about your question on #gf: you need ComplSlashPartLast from ExtendEng [15:09:57] [telegram] normal ComplSlash only does "squeeze in the cat" [15:09:57] [telegram] I will try it, thanks Inari! [15:09:57] [telegram] should I remove the Portuguese Extra module, as the Extend one is supposed to supersede it? [15:09:58] [telegram] (I figure there's no problem of backwards compatibility for Portuguese) [15:09:59] [telegram] Yeah go ahead and remove it! [15:10:00] [telegram] 👍 alright! I'll do some work on the Extend first, though! [15:10:05] [telegram] from what I understand, I'd say that as ComplSlashPartLast is English-specific, shouldn't we include it in the functor as ComplSlashPartLast = mkVP and let English handle it differently while everyone else uses the Functor? [15:10:05] [telegram] Germanic-specific more like, but yeah just regular ComplSlash in the functor sounds good [15:10:05] [telegram] ok! [15:10:10] [telegram] to add mkVP I must open the Syntax module, so I have to change all instances of Extend for all languages, right? (just so I don't do all this work if there's a better way!) [15:10:10] [telegram] phew, it works!! [15:10:10] [telegram] Matrix> p -lang=Eng "Browne squeezed the cat in" | l -lang=Por [15:10:12] [telegram] [PredVPS] [15:10:13] [telegram] Bobi apertava o gato [15:10:17] [telegram] You don't need to add mkVP from Syntax, just ComplSlash from Grammar (or Verb in particular, but it's included in Grammar) [15:10:19] [telegram] Sorry, should've been more specific yesterday! Is that the reason your pull requests fail? [15:10:21] [telegram] Or is it that ExtraPor is mentioned somewhere in a makefile or something? [15:10:23] [telegram] (what, *read* the build log? that's beyond my precious time! 😄) [15:10:26] [telegram] Btw, we discovered an embarrassing bug in gftest yesterday, I'll start fixing it and then the contexts generated should be much less redundant! [15:10:27] [telegram] great! I'll do it! [15:10:31] [telegram] hum, not sure, I forgot to check.. I'll do it soon! (re @Inari:You don't need to add mkVP from Syntax, just ComplSlash from Grammar (or Verb in particular, but it's included in Grammar)) [15:10:32] [telegram] that's great! catching digital bugs is awesome 😁 [15:10:35] [telegram] yesterday I thought I was online on IRC the whole day, but my connection dropped at some point and I didn't get a warning 😶 [15:11:32] *** Joins: e3928a3bc (b3d22a5d@gateway/web/freenode/ip.179.210.42.93) [15:11:56] hello! [15:12:17] hello! [15:13:49] [telegram] I'm having trouble trying to deprecate Extra* [15:13:50] [telegram] some of its funs are used in Construction* (and not all of them exist in Extend) [15:14:37] oh okay, I didn't know that! I would've thought everything relevant in Extra and Extensions is moved to Extend [15:14:41] [telegram] for instance, GenNP is used in ConstructionEng and exists in Extend, but this is not the case for ComplCN [15:14:46] it may just be an error or aarne didn't remember [15:15:05] [telegram] or maybe the names were changed? [15:15:30] are you trying to remove Extra from everything or just portuguese? [15:16:00] [telegram] it would be nice to start using Extend in Constructions, so that we don't have to maintain two equal functions in different places! (maybe import Extend in Extra to use all functions that should exist in both) [15:16:14] [telegram] only Portuguese, but I'm checking some others for guidance haha [15:17:36] yeah having same functions in two places doesn't sound ideal; defining Extra as Extend ** { functions not in Extend } sounds reasonable to me [15:17:49] you could ask it on the mailing list, in case other people have different opinions [15:18:35] don't think there's anything magical, just that people's GF application grammars don't break if they have imported Extra [15:31:52] [telegram] ok, will do! [15:32:02] [telegram] yes, we don't want that.. [15:32:31] [telegram] should I mention the ComplCN thing? [15:35:31] go ahead [15:36:08] it's good to have a concrete example [16:10:14] it's so exciting to generate tests for new languages, now I get "you are an uncertain husband" and "let's be wide queens" [16:11:16] "you are a water married to itself" sounds almost poetic [16:12:18] I should start a band and let the program generate all my song titles [16:13:59] [telegram] these are awesome! hahaha [16:14:03] [telegram] this is gftest? [16:14:10] yes [16:14:16] I'm testing the basque grammar [16:14:21] well, testing the test program after fixing a bug [16:14:30] [telegram] cool! [16:15:25] [telegram] I'm wondering why sometimes I find a variants {} where I'd expect a nonExist` -- what is the difference between them? [16:15:41] variants {} is just empty string, nonExist causes a runtime exception [16:16:07] [telegram] ahh, that makes sense [16:16:15] [telegram] but why not use the empty string then? [16:16:42] sometimes an empty string in a linearisation causes problems, you can parse it into anything [16:17:00] I don't actually know if variants {} is really empty string, or just "no linearisation, don't print anything" [16:18:04] if you have an actual, say, lin today_Adv = "" ; and AdvVP, AdvNP, AdvS etc, then you can say "dog" and also get the parse AdvNP (MassNP (UseN dog_N)) today_Adv [16:18:31] hmm I should test if it happens too with lin today_Adv = variants {} [16:18:51] or if you want to give it a try, feel free [16:18:57] I should be doing this thing I'm doing now [16:20:13] [telegram] yeah, it would make sense for it not to cause a lot of spurious ambiguities... [16:21:38] [telegram] inariksit: thanks! [20:28:01] [telegram] it seems that this will print things like [PredVPS] instead of giving runtime errors [20:28:01] [telegram] but how it does so, I'm not sure (re @ConcreteBot:[irc] I don't actually know if variants {} is really empty string, or just "no linearisation, don't print anything") [20:30:09] [telegram] as an example I tried to parse "she has slept" in English and in Portuguese. the extend functor implements PredVPS as variants {}, but English overrides it while Portuguese doesn't. [23:03:23] *** Quits: drbean (~drbean@TC210-63-209-169.static.apol.com.tw) (Read error: Connection reset by peer) [23:03:38] *** Joins: drbean (~drbean@210.60.168.196)