10:46 <discocaml> <mr.graa> hi can someone help me with this pls ?

10:46 <discocaml> <mr.graa> https://cdn.discordapp.com/attachments/436568060288172044/1221409717407191040/test.ml_-_.vscode_-_Visual_Studio_Code_3_24_2024_11_45_25_AM.png?ex=66127971&is=66000471&hm=624c8bc7f3a6bc810a4c1daac255e88dfbb451e9c7b908bfbd49e4ed03764767&

12:15 <Anarchos> hwo to build a dune version against a local ocaml compiler (5.3.0+dev0-2023-12-22)

13:55 <discocaml> <JM> I'm not sure I understand what problem you are facing

14:54 <Anarchos> discocaml i have a local repo with ocaml source (trunk + local modifications). I want a local repo to compile dune against this ocaml compiler

15:38 <Anarchos> discocaml but 'make dev' in dune complains about lwt-unix.

15:50 <discocaml> <Kali> discocaml is the name of the bridge

15:51 <Anarchos> JM i have a local repo with ocaml source (trunk + local modifications). I want a local repo to compile dune against this ocaml compiler

15:51 <Anarchos> Kali sorry for the naming...

17:11 <discocaml> <functionalprogramming> I want to use GADTs to create a list of "heterogeneous" functions that can be composed together: http://sprunge.us/pjLX4d. But I get a type error, and I think what I want is for in `apply`, it should be `forall a. exists b. ...` in the type signature. But IDK how to do that

17:16 <discocaml> <Kali> https://drup.github.io/2016/08/02/difflists/

17:17 <discocaml> <Kali> well, maybe that's not what you're looking for, but it may help

17:19 <discocaml> <functionalprogramming> okay yeah i am currently trying with 2 type variables

17:21 <discocaml> <functionalprogramming> i think i got it

17:23 <discocaml> <functionalprogramming> http://sprunge.us/5Q6bBt

17:23 <discocaml> <functionalprogramming> this look good? @ilo_kali

17:23 <discocaml> <functionalprogramming> needed two type variables

17:25 <discocaml> <Kali> seems good

17:25 <discocaml> <Kali> i would use [] and (::) rather than Id and Compose so that you can use the list syntax but otherwise looks fine if the compiler accepts it

17:27 <discocaml> <Kali> then instead of `Compose (Float.to_string, Compose (String.length, Compose ((fun x -> x * 4), Id)))` you can do `[Float.to_string; String.length; (*) 4]`

17:28 <discocaml> <Kali> (the `fun x -> 4 * x` can be shrunk to `(*) 4` either way)

17:28 <discocaml> <functionalprogramming> sweeeeet

17:29 <discocaml> <functionalprogramming> yep i did it

17:29 <discocaml> <functionalprogramming> thanks, it looks super neat

17:55 <discocaml> <functionalprogramming> is there any benefit to using this over just, compose : (a -> b) -> (b -> c) -> (a -> c) or smth like that @ilo_kali

17:59 <discocaml> <functionalprogramming> i guess u can remove functions which u cant do with compose

18:33 <discocaml> <octachron> You can replay the function chain, or apply it step by step.

18:35 <discocaml> <octachron> But the use with just functions are limited (at least without using the polymorphic equality)

18:36 <discocaml> <octachron> The technique is more useful with applicative with more behaviors, or with function-like constructors that you can destructure.

21:53 <discocaml> <functionalprogramming> can you give a brief example?