Newest at the top
| 2024-11-19 23:56:48 +0100 | L29Ah | (~L29Ah@wikipedia/L29Ah) (Error from remote client) |
| 2024-11-19 23:55:14 +0100 | <hellwolf> | you can do a lot with this trick... this needn't to be a amazement if GHC didn't have that "stage restriction". But for now, it's wonderful. |
| 2024-11-19 23:52:28 +0100 | <hellwolf> | I understood that I could skip the $() at top-level. |
| 2024-11-19 23:52:17 +0100 | <hellwolf> | but then... it seems all very reasonable that I didn't have to name it. |
| 2024-11-19 23:51:55 +0100 | <hellwolf> | I took your example and named it, and it didn't work. |
| 2024-11-19 23:51:38 +0100 | <hellwolf> | I didn't know the trick of not naming it. |
| 2024-11-19 23:50:59 +0100 | <Leary> | hellwolf: It is a `Q [Dec]`, implicitly wrapped in $(). The key to avoiding the stage restriction is to inline it into the top-level instead of naming it or its components. |
| 2024-11-19 23:50:25 +0100 | Tuplanolla | (~Tuplanoll@91-159-69-59.elisa-laajakaista.fi) Tuplanolla |
| 2024-11-19 23:48:18 +0100 | <hellwolf> | https://play-haskell.tomsmeding.com/saved/BnEqXAOh |
| 2024-11-19 23:47:54 +0100 | <hellwolf> | Thank you! |
| 2024-11-19 23:47:50 +0100 | <hellwolf> | https://play-haskell.tomsmeding.com/saved/fFgkdYeB <-- I saved the excerpt. |
| 2024-11-19 23:47:37 +0100 | haritz | (~hrtz@user/haritz) haritz |
| 2024-11-19 23:47:37 +0100 | haritz | (~hrtz@82-69-11-11.dsl.in-addr.zen.co.uk) (Changing host) |
| 2024-11-19 23:47:34 +0100 | haritz | (~hrtz@82-69-11-11.dsl.in-addr.zen.co.uk) |
| 2024-11-19 23:47:34 +0100 | tromp | (~textual@92-110-219-57.cable.dynamic.v4.ziggo.nl) (Quit: My iMac has gone to sleep. ZZZzzz…) |
| 2024-11-19 23:47:18 +0100 | Alleria_ | (~Alleria@user/alleria) (Ping timeout: 276 seconds) |
| 2024-11-19 23:44:40 +0100 | <hellwolf> | this is magic |
| 2024-11-19 23:44:34 +0100 | <hellwolf> | hallelujah |
| 2024-11-19 23:44:14 +0100 | <hellwolf> | and it did work! |
| 2024-11-19 23:44:12 +0100 | <hellwolf> | oh wait, so you didn't wrap it in a Q Dec |
| 2024-11-19 23:44:01 +0100 | Alleria | (~Alleria@user/alleria) Alleria |
| 2024-11-19 23:43:30 +0100 | haritz | (~hrtz@user/haritz) (Ping timeout: 252 seconds) |
| 2024-11-19 23:43:10 +0100 | <Leary> | hellwolf: https://play-haskell.tomsmeding.com/saved/T1DSbH8S |
| 2024-11-19 23:42:12 +0100 | <hellwolf> | (without full understanding of why) |
| 2024-11-19 23:41:58 +0100 | <hellwolf> | my dogmatic belief was that I should always use TH Qs imported. |
| 2024-11-19 23:41:34 +0100 | ljdarj1 | ljdarj |
| 2024-11-19 23:41:34 +0100 | ljdarj | (~Thunderbi@user/ljdarj) (Ping timeout: 260 seconds) |
| 2024-11-19 23:41:12 +0100 | <hellwolf> | me? I was expecting magic. |
| 2024-11-19 23:40:49 +0100 | <hellwolf> | but then it would warn about orphaned instances, which is probably not too bad. |
| 2024-11-19 23:40:23 +0100 | <geekosaur> | ? |
| 2024-11-19 23:40:22 +0100 | <hellwolf> | but I can indeed use that in a separate module, which doesn't make the module organization awkward |
| 2024-11-19 23:40:22 +0100 | <geekosaur> | what were you expecting? that ghc would pull the definition out and build it separately |
| 2024-11-19 23:39:42 +0100 | <hellwolf> | :( |
| 2024-11-19 23:39:41 +0100 | <hellwolf> | and must be imported, not defined locally |
| 2024-11-19 23:39:41 +0100 | <hellwolf> | ‘declare_all_valid_intn’ is used in a top-level splice, quasi-quote, or annotation, |
| 2024-11-19 23:39:41 +0100 | <hellwolf> | GHC stage restriction: |
| 2024-11-19 23:39:11 +0100 | ljdarj1 | (~Thunderbi@user/ljdarj) ljdarj |
| 2024-11-19 23:36:42 +0100 | mange | (~user@user/mange) mange |
| 2024-11-19 23:35:05 +0100 | <hellwolf> | (say no more, let me try, learning something every day) |
| 2024-11-19 23:34:12 +0100 | <Leary> | Non-trivial TH does, so that the functions you invoke within are already compiled. |
| 2024-11-19 23:34:04 +0100 | evocatus | (~evocatus@2a02:a210:20c2:d600:1496:234e:dc9f:5868) (Ping timeout: 272 seconds) |
| 2024-11-19 23:33:51 +0100 | esph | (~weechat@user/esph) esph |
| 2024-11-19 23:32:24 +0100 | <hellwolf> | I never truly understood why TH has to be in a separate module. |
| 2024-11-19 23:32:03 +0100 | <hellwolf> | oh?! was that a dogma in that I believed? |
| 2024-11-19 23:31:09 +0100 | <Leary> | Oh dear. Well, in any case, TH doesn't necessarily require another module. Try `flip foldMap [1..32] \i -> [d| instance ValidINTn $(litT (numTyLit i)) |]`. |
| 2024-11-19 23:28:31 +0100 | michalz | (~michalz@185.246.207.193) (Remote host closed the connection) |
| 2024-11-19 23:24:42 +0100 | <hellwolf> | https://pastebin.com/DqJ6yay3 <-- an excerpt, as of now. I am content with this version of ValidINTn, It restricts what "n :: Nat" a valid program can use, which is what I strive for. |
| 2024-11-19 23:24:39 +0100 | <hellwolf> | Understood. I would still prefer per instance scope. I guess it's for the granularity of controls. |
| 2024-11-19 23:22:23 +0100 | <Leary> | UndecidableInstances won't bite |
| 2024-11-19 23:21:06 +0100 | takuan | (~takuan@178-116-218-225.access.telenet.be) (Remote host closed the connection) |