[09:20:50] *** Joins: michmech (~Thunderbi@194.213.43.122) [09:21:22] *** Quits: michmech (~Thunderbi@194.213.43.122) (Client Quit) [09:52:46] daherb: maybe friday night? [12:13:55] inariksit: yes, that could work [12:36:03] let's keep that preliminarily and see what others think [13:33:03] *** Joins: michmech (~Thunderbi@194.213.43.122) [13:36:14] *** Quits: michmech (~Thunderbi@194.213.43.122) (Client Quit) [14:19:11] inariksit: a great initiative! i think friday night works for me. [14:40:22] okay! preferences for time, like 18, 19, 20, or later? [14:40:41] any preferences for platform? [14:45:18] any of those will work for me, i'd say. [14:46:34] as for platform, i don't have any strong opinions. do you have any preferences? [14:47:35] not really. I've been using discord and google hangouts in my previous video meetings, both have been working fine [14:48:56] haha not video from discord, just audio [14:50:25] i've been using zoom and blackboard collaborate (the latter for teaching purposes). zoom seems to have a bad rep atm for privacy, but it works fine. blackboard is fine if one wants to show slides, but otherwise not so great. [14:51:02] haha I think we can skip the slides from a beer [14:52:38] yes, i'm fine without the slides! [14:53:31] is google hangouts fine for you? I'll ask other people too what they prefer [14:53:43] hangouts is fine! [14:53:58] ok! [14:56:10] inariksit: i'm thinking of getting the api (partially) functional for yid; from you work on kor, it seems that i need some module like MissingYid with all the non-implemented stuff. is there anything else that needs to be done? [14:57:08] with the purpose of doing a rgl implementation of foods etc. [15:20:08] yes, that's right! [15:20:31] you can generate it with this https://inariksit.github.io/gf/2018/08/28/gf-gotchas.html#generating-missingxxx [15:21:02] so as long as you don't want yiddish to show in the synopsis, there's nothing else that needs to be done [15:22:50] technically you don't need a lot of the funs in Lexicon, but this method in my link doesn't look at all what's needed in Constructors/Combinators/etc. files [15:23:11] it doesn't do any harm if you have unnecessary opers in MissingYid, so that's why I didn't bother [15:24:34] thanks, great! [16:36:33] inariksit: i have good experience using jitsi. and its free and open source [18:34:25] a gf question: [18:35:48] at the moment i have overloaded paradigms: mkN, mkA, mkV ... in ResYid.gf, which create elements of types Noun, Adjective, Verb ... [18:36:15] and in CatYid i have lincat N = Noun etc. [18:36:37] now, when i'm trying to use the api, that does not seem to work. [18:37:42] should i just do it like many other languages, and move the public interface to ParadigmsYid instead? [20:34:11] can you be more specific what doesn't work? [20:34:23] ParadigmsYid is expected if you want your language to show up in the synopsis [20:34:45] you know, these descriptions https://www.grammaticalframework.org/lib/doc/synopsis/index.html#toc87 [20:35:05] the correct syntax to generate those is also a bit of an unknown folklore :-D [21:04:10] well, what i mean, is that at the moment, in files like gf-contrib/tutorial/resource-foods/LexFoodsYid.gf [21:04:49] i can't do: 'pizza_N = mkN "pizza" ;' [21:05:20] but instead i have to do 'pizza_N = lin N (mkN "pizza") ;' [21:06:26] my guess is that this is because when i go through the api it is not apparent that the linearisation types "Noun" (from ResYid) and "N" from CatYid are the same. [21:07:24] in LexiconYid OTHOH i can do 'pizza_N = mkN "pizza" ;' without problems. [21:36:42] oh yes those functor thingies cause a lot of weird problems [21:36:58] but yes, what you said is correct: "it is not apparent that the linearisation types "Noun" (from ResYid) and "N" from CatYid are the same." [21:37:37] more specifically, it's this problem https://inariksit.github.io/gf/2018/05/25/subtyping-gf.html#lock-fields [21:43:09] esg: i just checked, in my latin grammar i have Noun in ResLat and lincat N = Noun in CatLat but define the functions in ParadigmsLat using N instead of Noun [21:49:25] yeah that's the standard, ParadigmsLat exports the mkX constructors and those return the lincat versions, not the oper versions [21:54:28] thanks! well that all makes sense. now i start to see the point of ParadigmsXxx (except for hiding internal opers to external users). [21:55:04] daherb: how complete is the latin grammar at this point in time? [21:56:09] esg: that's a very good question [21:57:13] still quite a lot of stuff missing [21:57:50] but i did some work on evaluating it and it didn't seem too bad when looking at corpus data [21:58:38] daherb: how did you do that evaluation? [21:59:50] i did some ud2gf magic to get a frequency list of syntactic functions from the ud treebank [22:00:45] the paper is under submission for conference postproceedings but i guess i could mail you a copy [22:00:54] yes please! [22:01:46] esg: what's your mail address? [23:20:53] *** Joins: ski (~ski@remote11.chalmers.se) [23:25:09] daherb: emil.skoldberg@nuigalway.ie