Newest at the top
| 2024-12-24 02:30:53 +0100 | <EvanR> | loonycyborg, a lot of those libraries do exactly what they intended and have no need to keep updating |
| 2024-12-24 02:30:51 +0100 | <loonycyborg> | but actual values could be got from command line and other places |
| 2024-12-24 02:30:34 +0100 | <loonycyborg> | each string has own type specified by app or module |
| 2024-12-24 02:30:15 +0100 | merijn | (~merijn@128-137-045-062.dynamic.caiway.nl) merijn |
| 2024-12-24 02:30:14 +0100 | <haskellbridge> | <thirdofmay18081814goya> loonycyborg: and what is the target type? |
| 2024-12-24 02:29:50 +0100 | <Leary> | loonycyborg: There are also packages on hackage for vector/array backed extensible records. As for version numbers, being <1 doesn't mean anything. |
| 2024-12-24 02:29:36 +0100 | <loonycyborg> | strings are determined by app or modules so it's pretty specific |
| 2024-12-24 02:28:41 +0100 | <haskellbridge> | <thirdofmay18081814goya> are these any strings or a specific set of strings? |
| 2024-12-24 02:28:31 +0100 | <loonycyborg> | but they all seem to be abandoned by devs with version numbers before 1.0 |
| 2024-12-24 02:28:03 +0100 | <loonycyborg> | HList seems to mostly fit |
| 2024-12-24 02:27:48 +0100 | <EvanR> | but was too slow |
| 2024-12-24 02:27:39 +0100 | <EvanR> | was going to suggest looking at how dependent map works, which is already a library |
| 2024-12-24 02:27:32 +0100 | <geekosaur> | there's packages on Hackage for that already |
| 2024-12-24 02:27:31 +0100 | <haskellbridge> | <thirdofmay18081814goya> how many constructors does your type have? |
| 2024-12-24 02:27:23 +0100 | geekosaur | wonders if this is just a dependent map |
| 2024-12-24 02:27:05 +0100 | <EvanR> | Symbols are already types and look like strings |
| 2024-12-24 02:26:43 +0100 | <loonycyborg> | current idea is to make mapping of strings to types at compile time then make some kind of structure that can be looked up by that string to get value of appropriate type |
| 2024-12-24 02:25:52 +0100 | <geekosaur> | you'll need to think through what exactly you're doing to make that decision |
| 2024-12-24 02:25:36 +0100 | <geekosaur> | can you do something like reflecting a type level list into a runtime Vector? |
| 2024-12-24 02:25:34 +0100 | <EvanR> | or reasoning metapattern |
| 2024-12-24 02:24:59 +0100 | <EvanR> | any reasoning pattern you use must be reflected in the type system |
| 2024-12-24 02:24:34 +0100 | <EvanR> | and (before all that) verify what you plan to do at runtime makes any damn sense, which is where it gets complicated |
| 2024-12-24 02:23:24 +0100 | __monty__ | (~toonn@user/toonn) (Quit: Lost terminal) |
| 2024-12-24 02:21:51 +0100 | weary-traveler | (~user@user/user363627) (Remote host closed the connection) |
| 2024-12-24 02:19:35 +0100 | merijn | (~merijn@128-137-045-062.dynamic.caiway.nl) (Ping timeout: 260 seconds) |
| 2024-12-24 02:19:01 +0100 | j1n37 | (~j1n37@user/j1n37) j1n37 |
| 2024-12-24 02:18:24 +0100 | <loonycyborg> | ye I want to build mapping at compile time but make runtime structures based on it |
| 2024-12-24 02:16:37 +0100 | <EvanR> | or so I thought |
| 2024-12-24 02:16:26 +0100 | <EvanR> | it's not at runtime |
| 2024-12-24 02:16:09 +0100 | <loonycyborg> | at least at runtime |
| 2024-12-24 02:15:57 +0100 | <loonycyborg> | ye but avoiding traversal of whole list would be nice |
| 2024-12-24 02:15:55 +0100 | j1n37 | (~j1n37@user/j1n37) (Read error: Connection reset by peer) |
| 2024-12-24 02:14:57 +0100 | <EvanR> | if you literally don't care about order, then a list still works |
| 2024-12-24 02:14:52 +0100 | merijn | (~merijn@128-137-045-062.dynamic.caiway.nl) merijn |
| 2024-12-24 02:14:21 +0100 | <EvanR> | but a list is probably the most realistic |
| 2024-12-24 02:14:07 +0100 | <EvanR> | if you're feeling nuts you might have success implementing a binary tree in the type system which serves the same purpose |
| 2024-12-24 02:14:03 +0100 | <loonycyborg> | at *compile time |
| 2024-12-24 02:13:56 +0100 | acidjnk_new | (~acidjnk@p200300d6e7283f25e9e9d221c37061e3.dip0.t-ipconnect.de) (Ping timeout: 264 seconds) |
| 2024-12-24 02:13:53 +0100 | son0p | (~ff@186.121.100.67) son0p |
| 2024-12-24 02:13:53 +0100 | <loonycyborg> | I was thinking to use array at runtime and calculate indices for it at runtime |
| 2024-12-24 02:13:30 +0100 | ljdarj | (~Thunderbi@user/ljdarj) (Ping timeout: 265 seconds) |
| 2024-12-24 02:13:20 +0100 | <EvanR> | similar to KnownNat |
| 2024-12-24 02:13:19 +0100 | <geekosaur> | also Haskell's type system isn't really up to things like hashmaps; the best it can do is linked lists |
| 2024-12-24 02:13:13 +0100 | <EvanR> | KnownSymbol lets you have the corresponding String at runtime |
| 2024-12-24 02:12:11 +0100 | poscat | (~poscat@user/poscat) poscat |
| 2024-12-24 02:11:58 +0100 | <geekosaur> | ("dependent types" = types that depend on runtime values, not just the other way around) |
| 2024-12-24 02:11:27 +0100 | <geekosaur> | you really want Idris or Agda; Haskell doesn't have a real dependent types story as yet |
| 2024-12-24 02:11:03 +0100 | <geekosaur> | KnownSymbol and singletons (blech) |
| 2024-12-24 02:10:37 +0100 | acidjnk_new3 | (~acidjnk@p200300d6e7283f41b013a429fe4b7686.dip0.t-ipconnect.de) |
| 2024-12-24 02:08:51 +0100 | poscat0x04 | (~poscat@user/poscat) (Quit: Bye) |