[02:02:05] *** Joins: odanoburu (uid302941@gateway/web/irccloud.com/x-ngfejstxszqlybva) [06:46:22] *** Quits: odanoburu (uid302941@gateway/web/irccloud.com/x-ngfejstxszqlybva) (Quit: Connection closed for inactivity) [14:17:27] *** Quits: daherb (~daherb@cse-principia.cse.chalmers.se) (Ping timeout: 252 seconds) [14:42:30] *** Joins: odanoburu (uid302941@gateway/web/irccloud.com/x-ibmsfmthaongxipd) [15:09:37] odanoburu: I changed the blog post in the part that you had a question about (and bit elsewhere); it was the glu &+ ing morphemes thing [15:09:45] that aarne did to make the finnish grammar more feasible [15:09:53] (also fixed the links!) [15:10:25] nice, I'll reread for any insights! [15:10:45] btw, I might have another grammar blowup example in hands haha [15:10:53] it's only the intro + the last bit about basque, the rest is unchanged [15:10:57] odanoburu: ooh good [15:11:00] tell me about it :-D [15:11:33] I'm working with the arabic and it's cool otherwise but the text and RTL stuff is pretty annoying [15:17:21] so, the grammar is the Por concrete of https://github.com/odanoburu/gf-matrix [15:18:33] if you uncomment the first two funs and lincat, there's an error which I haven't solved yet; but if you also uncomment the only _Predet, the compilation segfaults, and I see no reason for that to happen o.O [15:19:52] okay let me try! [15:20:36] so the gluing method you describe in the post is something I should try for Portuguese verbs, I think [15:21:23] but I don't like the idea of having the glue tokens in parsed and linearized strings, though -- do you preprocess them every time you want to do any of these things? [15:21:27] I just get MatrixPor.gf: [15:21:27] MatrixPor.gf:21-24: [15:21:27] Happened in linearization of SlashV3S [15:21:27] cannot infer type of constant mkVPSlash [15:21:27] MatrixPor.gf: [15:21:27] MatrixPor.gf:13-19: [15:21:28] Happened in linearization of NPPPart [15:21:29] cannot infer type of constant heavyNPpol [15:21:50] if you parse them with C runtime, you don't need to preprocess [15:22:54] in fact, even in the standard haskell-based GF shell, there's a command ca (clitic_analyse) [15:23:36] where you specify a list of suffixes or prefixes, and then it tries to insert a &+ between those affixes and the rest of the text [15:24:41] even after you uncommented whatever_Predet? [15:25:21] I can't reproduce it in my pc, only in my university's pc, so I'm at a loss [15:25:22] duh no I didn't do that [15:25:25] now it's compiling [15:25:35] - parsing MatrixPor.gf renaming type checking [15:26:09] okay I'll leave it there and get back to arabic :_D [15:26:10] hum, well remembered... "if you parse them with C runtime, you don't need to preprocess" [15:26:33] ok! [15:37:22] okay now my computer starts suffering [15:49:44] that's interesting, my uni's computer segfaults pretty fast [15:50:27] and don't its exact specs but it's some kind of iMac from circa 2012 [15:50:37] I don't know* [16:06:57] I have a brand new macbook pro [16:07:07] digital grammars is so fancy :-D [16:07:32] hahaha [16:08:04] I have no idea how these compare, but I'm guessing the pro is pretty pro right? [16:08:31] Inari, I have some doubts about the Extend grammar; you worked on it didn't you? [16:08:57] do you know what the comment at L23 refers to? https://github.com/GrammaticalFramework/gf-rgl/blob/master/src/abstract/Extend.gf#L19-L26 [16:09:22] is it the fact that some languages might not implement all functions? [16:10:04] I didn't work on the abstract syntax [16:10:39] but if I have to guess, I'd say that maybe the first three don't generalise as well in all languages [16:19:47] ah ok! so far everything I parsed that has a tree with GenNP also has a tree with GenModNP [16:19:49] thanks! [16:24:22] np, also not sure if it's legit :-D [16:24:29] now I'm off to secret business! [16:26:15] "np, also not sure if it's legit :-D" didn't catch that, sorry! what's np? [16:51:40] *** Joins: daherb (~daherb@cse-principia.cse.chalmers.se)