Newest at the top
| 2025-04-06 19:14:41 +0200 | <emojelly> | Type applications seem at least a bit cleaner than "gratuitous" arguments... |
| 2025-04-06 19:11:01 +0200 | <haskellbridge> | ... long message truncated: https://kf8nh.com/_heisenbridge/media/kf8nh.com/JYSfMNZqjiXzhkfTeJfRrGsX/9kwQ5Tzz6so (4 lines) |
| 2025-04-06 19:11:01 +0200 | <haskellbridge> | <hellwolf> options: |
| 2025-04-06 19:10:26 +0200 | <emojelly> | Yeah that's possible... I don't quite remember if I had to disambiguate both. |
| 2025-04-06 19:10:03 +0200 | <emojelly> | Yeah. I've been wondering about using proxies... but then I think I have to pass a Proxy instead? |
| 2025-04-06 19:09:51 +0200 | <c_wraith> | But you still should only need to disambiguate t itself somehow, not m1 m2 a1 or a2 |
| 2025-04-06 19:09:33 +0200 | <emojelly> | It's actually a type family from MonadTransControl (which is a relatively common class), but it's the same if I define it myself of course. |
| 2025-04-06 19:09:16 +0200 | <c_wraith> | well that makes it a ton more complicated. |
| 2025-04-06 19:09:06 +0200 | <emojelly> | it is, yeah! |
| 2025-04-06 19:08:24 +0200 | <c_wraith> | unless StT is a type family? |
| 2025-04-06 19:07:30 +0200 | <c_wraith> | with that particular class definition, all that you normally should need to disambiguate is t |
| 2025-04-06 19:06:42 +0200 | <emojelly> | All the information about the (t m1 a1) and (t m2 a2) are containing in (StT t a1) (StT t a2) respectively, that's pretty much their point. |
| 2025-04-06 19:05:55 +0200 | <emojelly> | Is there a more elegant way to do that, than passing arguments which are not actually used for anything? |
| 2025-04-06 19:05:37 +0200 | tzh | (~tzh@c-76-115-131-146.hsd1.or.comcast.net) |
| 2025-04-06 19:05:36 +0200 | <emojelly> | Those first two arguments, of type (t m1 a1) and (t m2 a2), are only there to disambiguate the type. Otherwise, ghc complains that all the "St t"s are ambiguous. |
| 2025-04-06 19:04:48 +0200 | <emojelly> | combineC :: forall m1 m2 a1 a2. (t m1 a1) -> (t m2 a2) -> Maybe (StT t a1) -> Maybe (StT t a2) -> StT t (Maybe a1, Maybe a2) |
| 2025-04-06 19:04:46 +0200 | <emojelly> | ... |
| 2025-04-06 19:04:44 +0200 | <emojelly> | class MonadTransControl t => MonadTransCombine t where |
| 2025-04-06 19:04:42 +0200 | <emojelly> | So I have the following function in my type class: |
| 2025-04-06 19:04:04 +0200 | emojelly | (~eselber_p@user/endojelly) endojelly |
| 2025-04-06 19:03:32 +0200 | __monty__ | (~toonn@user/toonn) toonn |
| 2025-04-06 19:01:31 +0200 | ss4 | (~wootehfoo@user/wootehfoot) (Read error: Connection reset by peer) |
| 2025-04-06 19:01:13 +0200 | wootehfoot | (~wootehfoo@user/wootehfoot) (Read error: Connection reset by peer) |
| 2025-04-06 19:00:54 +0200 | ss4 | (~wootehfoo@user/wootehfoot) wootehfoot |
| 2025-04-06 18:53:22 +0200 | tromp | (~textual@2001:1c00:3487:1b00:c873:d422:44c2:cc0c) (Quit: My iMac has gone to sleep. ZZZzzz…) |
| 2025-04-06 18:43:49 +0200 | <ski> | er .. i suppose, not with `0'. same issue, though |
| 2025-04-06 18:42:57 +0200 | <ski> | but we want the neutral element of the given group to be identified with `-1', not with `1', of the group of integers |
| 2025-04-06 18:42:06 +0200 | <ski> | hmm .. no, that doesn't quite work. coproduct would identify the neutral elements of the two groups |
| 2025-04-06 18:39:39 +0200 | <ski> | (note, arbitrary groups, not abelian ones) |
| 2025-04-06 18:39:11 +0200 | <ski> | i suppose you could take coproduct of the given group, and the group of integers |
| 2025-04-06 18:37:40 +0200 | __monty__ | (~toonn@user/toonn) (Quit: Lost terminal) |
| 2025-04-06 18:37:21 +0200 | <mauke> | and inverse x' = -x - 2? |
| 2025-04-06 18:37:19 +0200 | ski | idly ponders making this into a word group |
| 2025-04-06 18:36:05 +0200 | hattckory | (~hattckory@bras-base-toroon4524w-grc-30-70-27-118-207.dsl.bell.ca) |
| 2025-04-06 18:33:46 +0200 | <monochrom> | Spin-off exercise: Suppose x +++ y = x + y + 1. Prove that +++ is a monoid operation with -1 as the identity. >:) |
| 2025-04-06 18:29:21 +0200 | <ski> | it sounded like thirdofmay18081814goya wanted something a bit reminiscent of the caching part, but for every bind automatically |
| 2025-04-06 18:28:32 +0200 | <ski> | i recall doing a CGI monad, which had an `io :: (Show a,Read a) => IO a -> CGI a' operation, caching the result of `IO' operations, serializing them into the generated page (when a query was generated), so that the program could then be resumed, fastforwarded to the point where it left off, to continue on |
| 2025-04-06 18:28:06 +0200 | <monochrom> | s/is the/in the/ |
| 2025-04-06 18:27:50 +0200 | <monochrom> | Of course we know the right way is to hide the counter increment is the something else. |
| 2025-04-06 18:27:42 +0200 | <enikar> | let monads do their job :) |
| 2025-04-06 18:27:12 +0200 | <monochrom> | Not to endorse it, but presumably the monad also does something else, and the author hides a counter increment in bind as a proxy to count cost. |
| 2025-04-06 18:26:58 +0200 | amadaluzia | (~amadaluzi@user/amadaluzia) (Ping timeout: 244 seconds) |
| 2025-04-06 18:24:52 +0200 | <monochrom> | :) |
| 2025-04-06 18:24:51 +0200 | <enikar> | I don't understand the use case. |
| 2025-04-06 18:24:34 +0200 | <ski> | dubious reasons |
| 2025-04-06 18:24:19 +0200 | <enikar> | why one would want to count bind? |
| 2025-04-06 18:22:52 +0200 | toby-bro | (~toby-bro@user/toby-bro) toby-bro |
| 2025-04-06 18:20:01 +0200 | <monochrom> | This is what the model "monad/>>= is programmable semicolon" gets right. In C programming, do people even imagine writing a program that counts its own lines of code? |
| 2025-04-06 18:15:45 +0200 | hattckory | (~hattckory@bras-base-toroon4524w-grc-30-70-27-118-207.dsl.bell.ca) (Ping timeout: 248 seconds) |
| 2025-04-06 18:15:22 +0200 | <monochrom> | But I don't like to tell people about this. Counting binds is every beginners favourite anti-monad anti-pattern. |