11:33 <neiluj> hi! have a library expose some types; but when used outside the type have the form Lib_Inner.t

11:33 <neiluj> instead of Lib.t

11:33 <neiluj> that's what merlin shows

11:33 <neiluj> is it normal?

11:34 <neiluj> if there is an open Lib, then t is equal to Lib_Inner.t

12:49 <masterbuilder> is the bridge down?

13:08 <companion_cube> Ah, maybe

13:13 <companion_cube> hmm

13:13 <discocaml> <companion_cube> bridge should be back

13:17 <discocaml> <Kali> oh, nevermind; i was confusing it with application, you are correct

13:34 <discocaml> <NULL> This is not associativity of *composition*, but of *application*, which is indeed false

13:42 <neiluj> given a variant type 'a t, it's not possible to match on the 'a right?

13:43 <neiluj> *polymorphic type

13:43 <discocaml> <NULL> Generally, you don't match on types

13:44 <neiluj> right

13:44 <neiluj> thx!

13:44 <discocaml> <NULL> If it's a GADT, you can do different stuff based on the type, but only as far as the constructors have different types and that you can match on those

13:45 <discocaml> <Kali> only if you match the surrounding type, then match the contents which contain `'a`

13:45 <discocaml> <Kali> if you had a `'b option option`, then to match on the `

13:45 <discocaml> <Kali> oops, i sent that too early

13:48 <discocaml> <Kali> if you had a `int option`, then to match on the value of `int`, you would first have to match on the option and get `Some ([value of int])`, otherwise you could not match on the `int` as there would be `None`

13:49 <discocaml> <Kali> unless you mean matching on the type itself

13:49 <discocaml> <Kali> not the value of the type

13:49 <discocaml> <Kali> in which case what i said is irrelevant

13:54 <neiluj> yeah I think what I want is to match on values, not types, thanks for the concrete example

13:56 <neiluj> trying to encode the relationships between some math objects: type _ t = Nat: gen -> nat t | Fp: gen -> ffield t | F: gen -> field t | Poly: gen -> ('a t) poly t

13:57 <neiluj> Two questions: 1. is it possible to refine the Poly constructor so that here 'a can either be nat, field or ffield?

13:57 <neiluj> 2. is it possible express "inclusions", ie. ffield t can be promoted to a field t

14:02 <octachron> `'a t poly t` means all `'a`.

14:02 <octachron> 2. Not in this way.

14:04 <neiluj> 2. can it be achieved without objects?

14:04 <octachron> You can use a common base constructor to be able to expression of the type `'a gen_field t`

14:04 <neiluj> oh thanks for the precision

14:05 <octachron> Objects at the type level matter mostly for subtyping, but since GADTs are not compatible with subtyping there are less helpful that one might think with GADTs.

14:07 <neiluj> ah gosh! so would you advocate the use of objects here? or GADT is fine

14:09 <octachron> I am not sure why you are thinking that I am suggesting using objects?

14:10 <neiluj> hmm I don't know either, I saw that for instance sagemath uses objects alot for subtyping

14:10 <neiluj> so was thinking maybe use this approach

14:12 <octachron> OCaml objects are not a good fit for small algebraic types.

14:12 <octachron> (too much method overhead)

14:14 <companion_cube> I mean, compared to python, it's still reasonable overhead... 😅

14:14 <neiluj> lol