00:08 <discocaml> <uberpyro181> thanks

01:49 <discocaml> <qrpnxz> Every time I change a public type I have to do it twice: first in the module, then again in the interface file.

01:49 <discocaml> <qrpnxz> Is there some way to tell OCaml that I want to export the whole type without having to play these games?

01:52 <discocaml> <functionalprogramming> _intf pattern

01:53 <discocaml> <functionalprogramming> create a module signature in a separate file, then use it in both the mli and ml files

01:53 <discocaml> <functionalprogramming> create a module signature in a separate file (xyz_intf.ml), then use it in both the mli and ml files

01:56 <discocaml> <yawaramin> that doesn't really solve this specific issue though

01:56 <discocaml> <qrpnxz> I reckon that's not gonna play well with ppx_deriving

01:57 <discocaml> <qrpnxz> I'd have to bring in ppx_import or smth

01:57 <discocaml> <qrpnxz> ugh

01:57 <discocaml> <deepspacejohn> Only expose it as an abstract type. (Probably not what you want but still)

01:57 <discocaml> <yawaramin> OCaml is designed with the intention that the type must be defined in both the interface and the implementation...you can try to bypass it with things like PPXs...but at the end of the imho it's not worth the extra complexity

01:58 <discocaml> <qrpnxz> It's just terrible friction while i'm trying to iterate

01:59 <discocaml> <deepspacejohn> I just don’t use mli files for modules that I’m doing a lot of iterations on

01:59 <discocaml> <qrpnxz> I'm thinking of straight up removing the interface, getting work done, then putting it back, hoping that i don't get namespace issues

01:59 <discocaml> <yawaramin> yeah if the module is in iteration mode it might not make sense to cover it up with an mli

02:00 <discocaml> <yawaramin> once the types have settled down and in maintenance mode, mli makes sense

02:00 <discocaml> <qrpnxz> yea, I think i'll go with that.

02:00 <discocaml> <qrpnxz> Thanks, all

02:01 <discocaml> <deepspacejohn> You can also use ppx_deriving annotations in mli files, FWIW. it will generate the signatures that go with the derived code in the ml

02:01 <discocaml> <qrpnxz> WHAT

02:01 <discocaml> <qrpnxz> That's gonna save me a lot of headache

02:01 <discocaml> <qrpnxz> have to try it

02:02 <discocaml> <qrpnxz> dang, it works. Awesome

02:02 <discocaml> <qrpnxz> bless you for that

02:02 <discocaml> <qrpnxz> I was removing them thinking it would try to put function in an interface or smth

02:02 <discocaml> <yawaramin> nice

02:02 <discocaml> <qrpnxz> lol

07:03 <discocaml> <lukstafi> Why? Just do the deriving in `xyz_intf.ml`, you get all the functions included. No?

07:08 <discocaml> <lukstafi> I'd start a new project without an interface to see how needs evolve. If it ends up being small types, then I add (copy-paste) `.mli` once it's settled what I want to advertise to the users or newcomers to the project. If it's large types like modules, I add (cut&paste) `_intf.ml`.

07:09 <discocaml> <lukstafi> (Ah, my comments were already made by others, sorry.)

11:31 <discocaml> <thecellist.> Hello, i'm currently learning about ocaml objects, and tried to implement a `fold` method, but apparently OCaml does not allow it to be polymorphic... 😔

11:34 <discocaml> <deepspacejohn> There's an example of a polymorphic fold in the manual: https://ocaml.org/manual/5.2/objectexamples.html#s:polymorphic-methods

11:34 <discocaml> <thecellist.> Here's my code btw

11:34 <discocaml> <thecellist.> https://cdn.discordapp.com/attachments/436568060288172044/1294261883628687401/image.png?ex=670a5ecf&is=67090d4f&hm=98ab585c687e6177a994754d3c4e570381b5c6c965f43d80293d462e922c753f&

11:34 <discocaml> <thecellist.> okay thanks

11:35 <discocaml> <deepspacejohn> TLDR you need to use an explicitly polymorphic annotation `method fold : 'a. ('a -> ...`

12:52 <discocaml> <thecellist.> oh okay i see

14:22 <discocaml> <mohamed_abdulsalam> ref🤮

14:35 <discocaml> <reynir1396> https://github.com/ocaml/ocaml/blob/b8f23dd5af09a3d9e6f891e2ac7f0d4baee81e1b/stdlib/array.ml#L188-L193

14:35 <discocaml> <mohamed_abdulsalam> heartbroken

15:25 <discocaml> <deepspacejohn> Imperative loops are completely reasonable times to use refs.

15:33 <discocaml> <leviroth> Look, let's be clear about what's going on here: if you look at that user's history in this channel they have eight or so messages, and at least two of them are rude remarks directed at different people for no particular reason.

15:35 <discocaml> <leviroth> I wouldn't want anyone to get the impression that this is a reasonable disagreement about code style. It's just that one person is going mean-spirited trolling, and the rest of us aren't.

15:35 <discocaml> <leviroth> I wouldn't want someone coming here for help to get the impression that this is a reasonable disagreement about code style. It's just that one person is going mean-spirited trolling, and the rest of us aren't.

17:15 <discocaml> <ada2k> the rest of us know what multi paradigm means i fear

19:04 <discocaml> <yawaramin> for the sake of completeness, https://github.com/scala/scala3/blob/5e7ab155c4cc117cdd78ef703056f4c5fdb84780/scala2-library-cc/src/scala/collection/IterableOnce.scala#L673