[07:18:14] *** Quits: koo5_ (~quassel@51.140.155.227) (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) [07:19:10] *** Joins: koo5_ (~quassel@51.140.155.227) [10:58:46] *** Joins: wmacmil (~wmacmil@c83-248-104-92.bredband.tele2.se) [11:00:42] does using pgf normally mess with the type inference abilities? [11:01:10] like i just am dubfounded why I can get a type-signature for this function [11:01:11] liftLinearize x = [11:01:11] do [11:01:11] gr' <- gr [11:01:11] eng' <- eng [11:01:11] let x' = gf x [11:01:13] return $ Atom (linearize gr' eng' x') [11:01:24] (where Atom :: String -> Phi) [11:02:25] like why when i include a partially completed type signature, liftLinearize :: _ -> IO Phi [11:02:39] it underlines the gf function saying [11:02:40] • Ambiguous type variable ‘a0’ arising from a use of ‘gf’ [11:02:40] prevents the constraint ‘(Gf a0)’ from being solved. [11:02:40] Relevant bindings include [11:03:07] and yet when i erase the type signature it infers something completely fine [11:05:28] and that it actually works polymorphically with respect to the GF categories, yet it fails to be callable in an outside function [11:06:50] i guess a big part of my confusion is the source of my confusion : like is it my failure to understand some simple monadic thing, some language extension nuance, or is this PGF in general [11:10:41] and why this failure to infer the signature also seems to render the liftLinearize function unsuable in any downstream function which calls it [11:10:53] i.e. -- callASTtoAtomDebug :: forall x. Tree x -> IO Phi [11:10:53] callASTtoAtomDebug x = case x of [11:10:53] GSimpleCom action -> liftLinearize action [11:10:53] _ -> return $ F (Atom "BASE") [11:10:53] [11:11:41] gives an inference failure with the use of liftLinearize, but when I include an explicit type signature, it doesn't linearize to the actual string [11:12:12] anyone more curious can see (https://github.com/wmacmil/commands2LTL/blob/master/app/InstrLTL.hs) [11:16:45] *** Quits: wmacmil (~wmacmil@c83-248-104-92.bredband.tele2.se) (Ping timeout: 256 seconds) [11:22:21] *** Joins: wmacmil (~wmacmil@83-233-165-97.cust.bredband2.com) [14:41:20] how does krasimir deal with (or other people) vocabulary holes the grammar doesn't cover when parsing an unseen expression that you still want to manipulate [14:48:31] you could ask that at the seminar on wednesday [14:49:00] *** ChanServ sets mode: +o daherb [14:50:21] *** daherb changes topic to 'Grammatical Framework | Next GF seminar February 9th 2022 http://www.grammaticalframework.org/~inari/gf-seminar/ | www.grammaticalframework.org | IRC logs at /irc' [14:50:44] *** ChanServ sets mode: -o daherb [14:58:59] wmacmil: in the gf shell one can use the -openclass parameter when parsing but not sure how to do it using the api [15:03:12] *** Quits: wmacmil (~wmacmil@83-233-165-97.cust.bredband2.com) (Quit: Leaving) [16:19:11] *** Joins: wmacmil (~wmacmil@83-233-165-97.cust.bredband2.com) [16:51:50] also is there a way to use template haskell to turn a string representing an ast to a pgf representation ? [16:51:57] tack herbert [18:40:09] *** Joins: Lycurgus (~juan@98.4.112.204) [20:37:52] *** Quits: inariksit (~inari@ksit.fixme.fi) (*.net *.split) [20:42:24] *** Joins: inariksit (~inari@ksit.fixme.fi) [22:35:47] *** Quits: Lycurgus (~juan@98.4.112.204) (Quit: Exeunt)