12:58 <Guest45> Hi! I'm debating between using higher-order functions, or use first-order modules for functions like `binary_exp ~mul x n` or `binary_exp (module G: Group) (x : G.t) (n : int) = ... G.mul x x ...` where the latter is nice from the user's perspective as they don't have to explicitly pass the operations they need, and make them comply with the

12:58 <Guest45> signature of mul

12:59 <Guest45> on the other hand, I wonder what is the added cost in terms of performance of passing the module as argument to `binary_exp`, compared to passing a pointer to a function

13:04 <Guest45> with the last option (first-order modules), the nice thing is that I can create functions that return a new module defining a new group at runtime, allowing for the creation of fresh types at runtime (G.t)

13:52 <discocaml> <gooby_clown> I don't think there's anything nice about first class modules from the user's perspective

14:40 <Guest97> gooby_clown: what makes you think so?

15:15 <discocaml> <gooby_clown> Overall cluttered/heavy syntax from what I've seen

15:16 <discocaml> <gooby_clown> And since OCaml doesn't have implicit modules you are still forced to pass *something*

15:17 <discocaml> <gooby_clown> Making that something a first class module doesn't seem to me like it would make anything easier to use, if anything potentially harder

15:19 <discocaml> <gooby_clown> In fact recently Jane Street has exercised its influence to make EIO use first class modules instead of objects, and in my opinion it strictly made the library less comfortable to use, although at the end of the day it's a personal choice

15:20 <Guest97> I see, thanks!

15:42 <Anarchos> first class modules , aren't they tag as experimental features ?

15:54 <discocaml> <deepspacejohn> I don’t think so. Recursive modules are still experimental though.

16:38 <discocaml> <sim642> I don't think recursive modules are experimental either, they've been around for 20 years now

16:38 <Guest97> omg my ocaml library uses a C library that allocates everything in a preallocated stack, that I now have to garbage collect manually...

16:39 <Guest97> i m gonna die

16:41 <discocaml> <octachron> It depends a lot of the meaning of experimental. Neither recursive modules nor first class modules will be ever removed.

16:42 <Anarchos> octachron sure, but they are still in "chapter 8 : language extensions" in the manual

16:43 <discocaml> <octachron> At the same time, no one would hesitate to apply some breaking changes (in corner cases) to make recursive modules fit a good specification.

16:45 <discocaml> <octachron> This section is mostly non-meaningful. It semantic is mostly: changes older than 3.12 that I have yet to convince other maintainers that it is high time to move them to the main part of the manual. (except for recursive modules, values, and effect handlers that are still mostly in flux).

16:50 <Anarchos> makes sense :)

16:51 <Anarchos> why do recursive modules have a non good specification ?

16:52 <discocaml> <octachron> They don't even have a specification, there is still many open question about then.

16:53 <octachron> Typically, there was a change in 5.1 to avoid an issue with classes inside recursive modules.

17:10 <Anarchos> yeah strange to see classes so powerful, and use so little

17:24 <Guest49> hi! is there a way to specify the number of iterations for Crowbar tests?

17:38 <Guest49> hmm, actually I need qcheck instead

23:08 <discocaml> <darrenldl> Guest49: you can pass a timeout argument to afl-fuzx

23:08 <discocaml> <darrenldl> afl-fuzz

23:09 <discocaml> <darrenldl> but crowbar itself has no control afaict

23:10 <discocaml> <darrenldl> (would be an odd design if program being fuzzed and stop the fuzzer)