2026/01/01

Newest at the top

2026-01-01 01:19:14 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2026-01-01 01:15:02 +0100 <janus> such a program, without interleaving, would have ambiguous types and compilation would fail. since it wouldn't be able to take the 'output' and propagate it to other constraints
2026-01-01 01:13:38 +0100 <janus> then, it should evaluate the fundep and that should give it enough info to find the 'principal' type
2026-01-01 01:13:17 +0100 <janus> maybe the easiest example should be something that uses mtl, and somehow has polymorphic types in both the 'input' and the 'output' type of a fundep. now, the surrounding program somehow determines one of these, and the typechecker should be able to run enough of 'regular' constraints to find out that there is this constraint on input and output
2026-01-01 01:10:36 +0100 <janus> i still can't imagine a program that would require something like that
2026-01-01 01:10:07 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 246 seconds)
2026-01-01 01:10:04 +0100 <janus> right. so it's not strictly MPTC that require interleaving
2026-01-01 01:09:38 +0100 <geekosaur> inference isn't possible otherwise
2026-01-01 01:09:24 +0100 <geekosaur> my understanding is that MPTCs are really only usable with either fundeps or TFs
2026-01-01 01:07:51 +0100 <janus> if i understand correctly, fundeps only make sense with MPTC, and since fundeps are so prevalent with mtl, i wonder whether that is why you said MPTC instead of saying fundeps. since for all practical typecheckers, they will need to support both
2026-01-01 01:06:47 +0100 <janus> [exa]: is it really just MPTC that would require type-level computation interleaved with previous typechecking? or is it about fundeps?
2026-01-01 01:05:18 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2026-01-01 01:03:55 +0100traxex(traxex@user/traxex) (Ping timeout: 264 seconds)
2026-01-01 00:57:15 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 240 seconds)
2026-01-01 00:55:55 +0100L29Ah(~L29Ah@wikipedia/L29Ah) ()
2026-01-01 00:52:38 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2026-01-01 00:44:11 +0100Everything(~Everythin@46.96.10.169) (Quit: leaving)
2026-01-01 00:43:49 +0100Pixi(~Pixi@user/pixi) (Ping timeout: 264 seconds)
2026-01-01 00:40:38 +0100Pixi`(~Pixi@user/pixi) Pixi
2026-01-01 00:32:23 +0100Pozyomka(~pyon@user/pyon) pyon
2026-01-01 00:25:34 +0100 <monochrom> Do a pair of complementary things and do them well: Open a pastry bakery and a gym next to each other. (True story.)
2026-01-01 00:24:58 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 246 seconds)
2026-01-01 00:18:17 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2026-01-01 00:16:30 +0100L29Ah(~L29Ah@wikipedia/L29Ah) L29Ah
2026-01-01 00:15:31 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 246 seconds)
2026-01-01 00:10:59 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2025-12-31 23:57:35 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 240 seconds)
2025-12-31 23:53:14 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2025-12-31 23:40:10 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 246 seconds)
2025-12-31 23:35:44 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2025-12-31 23:32:25 +0100peterbecich(~Thunderbi@71.84.33.135) peterbecich
2025-12-31 23:28:33 +0100wootehfoot(~wootehfoo@user/wootehfoot) (Read error: Connection reset by peer)
2025-12-31 23:23:48 +0100 <geekosaur> now I'm reminded of a TV ad which had a pizza delivery car roll up, then the pizza sign rotated into (IIRC) a real estate agent sign
2025-12-31 23:23:45 +0100L29Ah(~L29Ah@wikipedia/L29Ah) ()
2025-12-31 23:21:58 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 246 seconds)
2025-12-31 23:18:29 +0100rekahsoft(~rekahsoft@70.51.99.245) rekahsoft
2025-12-31 23:17:16 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2025-12-31 23:13:22 +0100 <EvanR> another one locally which should have had that billboard: this place "pool tables and pools"
2025-12-31 23:12:01 +0100 <EvanR> I think it was either albuquerque or atlanta, a billboard for "divorce lawyer and movers -- do one thing and do it well"
2025-12-31 22:53:16 +0100wickedjargon(~user@64.114.24.74) (Ping timeout: 246 seconds)
2025-12-31 22:33:39 +0100 <int-e> or two, or many
2025-12-31 22:33:33 +0100yotic(~yotic@user/yotic) yotic
2025-12-31 22:33:21 +0100 <int-e> that's fine because you can still have a tool for each thing
2025-12-31 22:24:57 +0100 <yin> the probem with the "one thing" philosophy is that "thing" is not well-defined
2025-12-31 22:22:49 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 244 seconds)
2025-12-31 22:16:15 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn
2025-12-31 22:12:25 +0100wennefer0(~wennefer0@user/wennefer0) (Ping timeout: 245 seconds)
2025-12-31 22:11:16 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 246 seconds)
2025-12-31 22:08:08 +0100wennefer0(~wennefer0@user/wennefer0) wennefer0
2025-12-31 22:06:55 +0100merijn(~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn