Newest at the top
| 2024-11-19 14:43:23 +0100 | <dminuoso> | 13:39:10 dminuoso │ bwe: i.e. `f :: HasFoo a => proxy a -> ...` |
| 2024-11-19 14:43:16 +0100 | <bwe> | oh, that's it. |
| 2024-11-19 14:43:01 +0100 | <dminuoso> | Always make it an argument. |
| 2024-11-19 14:42:46 +0100 | <dminuoso> | bwe: Just add an extra argument. |
| 2024-11-19 14:42:34 +0100 | <bwe> | dminuoso: I don't grasp how to use Proxy instead of Const. `Const Int a` becomes what using Proxy? |
| 2024-11-19 14:42:01 +0100 | <dminuoso> | Sort of how ZonedTime has no Eq instance because there's two different ways to think about equality, none of them clearly better. |
| 2024-11-19 14:41:40 +0100 | <[exa]> | dminuoso: yeah me too, which brings me to the question "why is the default `deriving Ord` THE authoritative one" |
| 2024-11-19 14:40:47 +0100 | <dminuoso> | [exa]: If there is not a single authoritative instance, I would hide them behind newtypes. |
| 2024-11-19 14:40:00 +0100 | <dminuoso> | bwe: (I have added the constraint just as an example, since many uses of Proxy usually only make sense together with a constraint) |
| 2024-11-19 14:39:53 +0100 | <hellwolf> | but it was fun to discover the N ways of doing the same thing. |
| 2024-11-19 14:39:52 +0100 | rvalue | (~rvalue@user/rvalue) rvalue |
| 2024-11-19 14:39:42 +0100 | <hellwolf> | Leary: in that case, I still think either has lower requirement of for Haskell knowledge for readability |
| 2024-11-19 14:39:35 +0100 | bitdex | (~bitdex@gateway/tor-sasl/bitdex) (Quit: = "") |
| 2024-11-19 14:39:31 +0100 | <[exa]> | Is there any "good natural" way to make an instances for `Ord (Tree a)` other than what is in Data.Tree? The one there is the automatic one obtained with `deriving`, i.e. basically follows the syntax. I feel like that ordering is very left-subtree-biased but no idea how to compensate for that (and esp. if there's some ground reason for why not). |
| 2024-11-19 14:39:29 +0100 | <dminuoso> | Which is essentially just `f :: forall proxy a. HasFoo a => proxy a -> ...` |
| 2024-11-19 14:39:10 +0100 | <dminuoso> | bwe: i.e. `f :: HasFoo a => proxy a -> ...` |
| 2024-11-19 14:38:44 +0100 | <dminuoso> | bwe: No, as a type variable. |
| 2024-11-19 14:38:41 +0100 | <Leary> | hellwolf: All approaches should optimise to pretty much the same code; it shouldn't matter. |
| 2024-11-19 14:38:38 +0100 | <bwe> | dminuoso: Do you mean I should define `proxy` as my own function? I don't find `proxy` in https://hackage.haskell.org/package/base-4.20.0.1/docs/Data-Proxy.html |
| 2024-11-19 14:38:37 +0100 | <hellwolf> | I mean, okay, what is the second example of proxy? from a differenr base, perhaps? |
| 2024-11-19 14:38:20 +0100 | <dminuoso> | hellwolf: It allows for arbitrary parameterized types. |
| 2024-11-19 14:38:05 +0100 | <hellwolf> | dminuoso: I never understood that. I see base code having lower case proxy in many places. why? |
| 2024-11-19 14:37:54 +0100 | L29Ah | (~L29Ah@wikipedia/L29Ah) () |
| 2024-11-19 14:37:15 +0100 | <dminuoso> | bwe: For extra points parameterize by `proxy Foo` rather than `Proxy `Foo` |
| 2024-11-19 14:37:07 +0100 | <hellwolf> | is Const a r Coercible with a? |
| 2024-11-19 14:36:45 +0100 | <dminuoso> | bwe: Yeah, just use Proxy. |
| 2024-11-19 14:36:24 +0100 | <dminuoso> | Leary: That `find (const True)` is cunning. :-) |
| 2024-11-19 14:36:20 +0100 | <bwe> | dminuoso: that's exactly where I am stuck right now: applying a function wrapped in a Const to values wrapped in Const |
| 2024-11-19 14:36:09 +0100 | <hellwolf> | Leary: I could balance between readability (which is subjective, up to the code reader's knowledge about all these instances) and performance; so what about any performance consideration? |
| 2024-11-19 14:35:48 +0100 | rvalue | (~rvalue@user/rvalue) (Ping timeout: 265 seconds) |
| 2024-11-19 14:35:23 +0100 | <dminuoso> | And it will probably require using ScopedTypeVariables and some annoyances. |
| 2024-11-19 14:35:11 +0100 | <dminuoso> | Const will constantly (the pun!) be in the way because you need to wrap/unwrap potentially many times |
| 2024-11-19 14:34:51 +0100 | <dminuoso> | bwe: Id say Proxy is the least akward and most common way. |
| 2024-11-19 14:34:21 +0100 | <Leary> | Personally, I would go with `asum`. Either way, get them from `Data.Foldable`; a `Data.List` import suggests list specialisation (which could actually happen at some point). |
| 2024-11-19 14:31:48 +0100 | <hellwolf> | import Data.List (find) is required though |
| 2024-11-19 14:31:38 +0100 | <hellwolf> | hard to choose :) |
| 2024-11-19 14:31:00 +0100 | <hellwolf> | either (const Nothing) (Just . ADDR) (S.decode b) |
| 2024-11-19 14:30:59 +0100 | <hellwolf> | or |
| 2024-11-19 14:30:53 +0100 | <hellwolf> | fmap ADDR . find (const True) $ S.decode b |
| 2024-11-19 14:28:28 +0100 | <hellwolf> | Left elements are mempty, right elements are actually "foldable" |
| 2024-11-19 14:28:05 +0100 | <hellwolf> | it all hinges all the fact of the Foldable Either instance. |
| 2024-11-19 14:27:46 +0100 | <hellwolf> | that's very cute, indeed |
| 2024-11-19 14:19:34 +0100 | housemate | (~housemate@2a04:9dc0:0:162::5d91:d7ed) housemate |
| 2024-11-19 14:10:18 +0100 | housemate | (~housemate@2a04:9dc0:0:162::5d91:d7ed) (Ping timeout: 244 seconds) |
| 2024-11-19 14:07:17 +0100 | <Leary> | hellwolf: ^ |
| 2024-11-19 14:07:14 +0100 | <yahb2> | find \_ -> True :: Foldable t => t a -> Maybe a |
| 2024-11-19 14:07:14 +0100 | <Leary> | % :t find \_ -> True |
| 2024-11-19 14:07:08 +0100 | <Leary> | Oh, bit late, but I thought of something cute and silly! |
| 2024-11-19 14:05:37 +0100 | sand-witch | (~m-mzmz6l@vmi833741.contaboserver.net) |
| 2024-11-19 14:01:28 +0100 | ash3en | (~Thunderbi@2a03:7846:b6eb:101:93ac:a90a:da67:f207) (Remote host closed the connection) |