05:42 <d_bot> <Bluddy> #editor-support should really be under Ecosystem. FYI I'll be moving it soon.

09:39 <d_bot> <andreypopp> yeah... looks like this isn't possible — caml_startup is only called for the first .so loaded. Wondering if ocaml 5 allows that

11:22 <d_bot> <Continuation Calculus> for testing purposes, I'm trying to concretize a virtual library's signature, and I can't even get define the type signature I want. `module type TYPE = module type of My_virtual_library with My_param.t = int` doesn't work (type declarations do not match,)

11:22 <d_bot> <Continuation Calculus> It looks like as if `My_param.t` already was concrete, but at the same time, I can't find what its concrete type is

11:25 <d_bot> <Continuation Calculus> Something that might help, how can I ask OCaml what all the constraints it has on a type are?

11:29 <d_bot> <Continuation Calculus> Or even better, given this is only for testing purposes, and not the actual live code, is there an `Obj.magic` at the module type substitution level, where I can be like "no worries, they really are the same type, let me perform the substitution"? I know how to create fake type equality witnesses through GADT and Obj.magic, but not sure how to use this to actually perform the `with type =` that I want

11:32 <d_bot> <octachron> `with` constraints are nearly textual transformation. You cannot have `Obj.magic` at the module level, because it would potentially break the typing of your programs rather than its execution.

11:33 <d_bot> <Continuation Calculus> > because it would potentially break the typing of your program

11:34 <d_bot> <octachron> I would suggest to just check the inferred type of `module type of My_virtual_library` (with `cmi2mli` for instance)

11:35 <d_bot> <octachron> Not really, you can create proofs of false, but your theorems are still well-written theorems (that states non-sensical results).

11:38 <d_bot> <octachron> Or in other words, the typechecker implementation is not written defensively against non-valid module types (or types) and expects well-formed ASTs.

11:42 <d_bot> <octachron> (And to explain the restriction on the `with` constraint, the issue would be with things like https://gist.github.com/Octachron/ce055a497cc8f1cdf03f8c252487599c that try to replace an injective type constructor with a non-injective one in a signature that relied on the injectivity of the original type constructor)

11:43 <d_bot> <Continuation Calculus> I see, thanks

11:58 <d_bot> <Continuation Calculus> .cmis are only built for libs, not bins? 😦

12:00 <d_bot> <octachron> Ah, cmi are built for everything except potentially the main module of an executable.

12:00 <d_bot> <Continuation Calculus> I see

12:01 <d_bot> <Continuation Calculus> I've built a dummy library, but only get a .a and a .cmxa

12:02 <d_bot> <octachron> The `cmi`s must be elsewhere, you cannot use library (or any compilation unit really) without `cmi`s

12:04 <d_bot> <Continuation Calculus> well, it's not in `$cur__target_dir/default/..path../`

12:05 <d_bot> <Continuation Calculus> I'll check out later

12:05 <d_bot> <Continuation Calculus> thanks tho, didn't know about cmitomli, epic tool

15:44 <sleepydog> the Fmt package is so good. I am constantly surprised by how succinctly I can define a pretty-printer for complex structures with this library.

18:49 <companion_cube> @octachron on a scale of "nope" to "lolnope", how likely is it that `let@ f x = f x` could make it into Fun?

18:49 <companion_cube> after using it a bit, it's just amazing, and I can't really imagine go back to `foo @@ fun x ->\n…`

19:03 <octachron> https://github.com/ocaml/ocaml/pull/9887 had mixed reactions for introducing `let@` in the compiler code base.

19:04 <octachron> I am not sure how those reactions translate to the stdlib context however

19:05 <companion_cube> forgot this issue, thank you for the link

19:05 <companion_cube> (although I did subscribe to it back then :D)

19:58 <d_bot> <glennsl> What's a "migration error" , as emitted by ocamlc?

19:58 <d_bot> <glennsl> More specifically `migration error: existentials in pattern-matching is not supported before OCaml 4.13`

19:58 <d_bot> <glennsl> I'm using 4.1.3.1

19:59 <d_bot> <glennsl> Oh, it's probably a PPX isn't it...

20:04 <d_bot> <EduardoRFS> yup

20:05 <d_bot> <EduardoRFS> I think ppxlib is on OCaml 4.12 AST

20:09 <d_bot> <glennsl> Hmm, this seems to suggest it supports 4.14 even: https://github.com/ocaml-ppx/ppxlib/blob/0.24.0/ast/supported_version/supported_version.ml

20:09 <d_bot> <glennsl> For the version of ppxlib I have installed

20:15 <sim642> But it currently migrates everything to 4.12 or something

20:22 <d_bot> <glennsl> Right, makes sense. I find it unreasonably hard to find that important piece of information though, but it's in the change log for 0.22, released a bit over a year ago: "Bump ppxlib's AST to 4.12 "

20:25 <d_bot> <glennsl> Looks like 4.14 AST is inbound very soon too: https://github.com/ocaml-ppx/ppxlib/pull/320

20:25 <d_bot> <glennsl> But might take a while for some PPXes to migrate as there's breaking changes

22:16 <d_bot> <AZAH> ```ml

22:16 <d_bot> <AZAH> val concat_map : ('a -> 'b list) -> 'a list -> 'b list

22:16 <d_bot> <AZAH> ```

22:16 <d_bot> <AZAH> concat_map f l gives the same result as List.concat (List.map f l). Tail-recursive.

22:16 <d_bot> <AZAH>

22:16 <d_bot> <AZAH> ```ml

22:16 <d_bot> <AZAH> val concat : 'a list list -> 'a list

22:16 <d_bot> <AZAH> ```

22:16 <d_bot> <AZAH>

22:16 <d_bot> <AZAH> How can concat_map give the same result as concat (map) when it expects 'a list list

23:22 <pgiarrusso> AZAH: since `f : 'a -> 'b list`, `map f xs: 'a list list` (for the right `xs`), and `concat` gets back to `'a list`.

23:24 <d_bot> <NULL> Are you answering the question from # beginners ?