OCaml Weekly News

Previous Week Up Next Week


Here is the latest OCaml Weekly News, for the week of May 31 to June 07, 2022.

Table of Contents


Deep in this threas, Fabian said

Note that you can, to a certain degree, build your own flat structures with the Bytes module. Compared to bigarrays, Bytes.t has less indirection, a lower constant memory overhead and can be allocated on the minor heap. The contents of Bytes.t are not scanned by the GC, just like bigarrays.

For example, a more efficient int32 Array.t:

module Int32_array : sig
  type t
  val equal : t -> t -> bool
  val create : int -> t
  val length : t -> int
  val get : t -> int -> int32
  val set : t -> int -> int32 -> unit
  val sub : t -> int -> int -> t
  val to_list : bytes -> int32 list
end = struct
  type t = Bytes.t
  let equal = Bytes.equal
  let create len = Bytes.create (4 * len)
  let length t = Bytes.length t / 4
  let get t i = Bytes.get_int32_le t (4 * i)
  let set t i x = Bytes.set_int32_le t (4 * i) x
  let sub t pos len = Bytes.sub t (4 * pos) (4 * len)
  let to_list t = List.init (length t) (get t)

A more efficient (int * int):

module Point : sig
  type t
  val create : int -> int -> t
  val x : t -> int
  val y : t -> int
end = struct
  external get_int64_unsafe : bytes -> int -> int64 = "%caml_bytes_get64u"
  external set_int64_unsafe : bytes -> int -> int64 -> unit = "%caml_bytes_set64u"
  type t = Bytes.t
  let create x y =
    let p = Bytes.create 16 in
    set_int64_unsafe p 0 (Int64.of_int x);
    set_int64_unsafe p 8 (Int64.of_int y);
  let x t = Int64.to_int (get_int64_unsafe t 0)
  let y t = Int64.to_int (get_int64_unsafe t 8)

(making a more efficient (int * int) Array.t is left as an exercise to the reader)

The downside compared to bigarrays is that it doesn't support sub without copying. Also, bytes can be moved by the GC (during minor GCs or compaction), and therefore you cannot release the runtime lock when passing them to C. The latter point is less relevant with the multicore extensions, especially since there is no compactor yet. There is some related discussion on the eio repository: https://github.com/ocaml-multicore/eio/issues/140

ML Family Workshop 2022: Final Call for Presentations

Benoit Montagu announced

ML Family Workshop 2022: DEADLINE EXTENSION

To increase your chances of submitting your work to the ML workshop, the submission deadline is extended by a week. The new deadline is Friday 10th June (AoE).

A quick reminder:

OCaml Users and Developers Workshop 2022

Matija Pretnar announced

To offer additional opportunities to contribute to the OCaml workshop, and to align with the ML family workshop, to which you are also cordially invited, the submission deadline has been extended by a week to Friday, June 10 (anywhere on Earth).

dkml-c-probe.2.0.0: Cross-compiler friendly definitions for C compiling

jbeckford announced

Summary: dkml-c-probe is a new package for maintainers who compile or link C code. Install it with opam install dkml-c-probe. Full docs are at https://github.com/diskuv/dkml-c-probe#readme


You are creating an OCaml package that has foreign C code. Perhaps you need special C headers or libraries when you are targeting Apple users, or perhaps you need to execute custom OCaml code for Android users. More generally you need a way to determine whether your OCaml or C code is compiling for a Linux AMD/Intel 64-bit, Android ARM 32-bit, or any other ABI target.


A user of your OCaml package may, for example, be on a 64-bit AMD/Intel Linux machine using a 32-bit OCaml system compiled with gcc -m32; additionally they have a 32-bit Android ARM cross-compiler toolchain. dkml-c-probe will tell you the target operating system is Linux and the target ABI is Linux_x86 except when the cross-compiler toolchain is invoked. With the cross-compiler toolchain dkml-c-probe will tell you the target operating system is Android and the target ABI is Android_arm32v7a.

How it works

dkml-c-probe uses C preprocessor definitions (ex. #if TARGET_CPU_X86_64, #if __ANDROID__, etc.) to determine which ABI the C compiler (ex. ocamlopt -config | grep native_c_compiler) is targeting.

This isn't a new idea. The pattern is used in Esy and Mirage code as well. dkml-c-probe just codifies the pattern for use in your own code.


In OCaml code you can use the versioned module:

module V2 :
    type t_os = Android | IOS | Linux | OSX | Windows
    type t_abi =
      | Android_arm32v7a
      | Android_x86
      | Android_x86_64
      | Darwin_arm64
      | Darwin_x86_64
      | Linux_arm64
      | Linux_arm32v6
      | Linux_arm32v7
      | Linux_x86_64
      | Linux_x86
      | Windows_x86_64
      | Windows_x86
      | Windows_arm64
      | Windows_arm32
    val get_os : (t_os, Rresult.R.msg) result Lazy.t
    val get_abi : (t_abi, Rresult.R.msg) result Lazy.t
    val get_abi_name : (string, Rresult.R.msg) result Lazy.t

In C code you can use the provided dkml_compiler_probe.h header from within Dune or Opam. Here is a snippet that handles part of the Linux introspection:

#elif __linux__
#   if __ANDROID__
#       ...
#   else
#       define DKML_OS_NAME "Linux"
#       define DKML_OS_Linux
#       if __aarch64__
#           define DKML_ABI "linux_arm64"
#           define DKML_ABI_linux_arm64
#       elif __arm__
#           if defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) ||
defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6Z__) || defined(__ARM_ARCH_6ZK__) ||
#               define DKML_ABI "linux_arm32v6"
#               define DKML_ABI_linux_arm32v6
#           elif defined(__ARM_ARCH_7__) || defined(__ARM_ARCH_7A__) ||
defined(__ARM_ARCH_7R__) || defined(__ARM_ARCH_7M__) || defined(__ARM_ARCH_7S__)
#               define DKML_ABI "linux_arm32v7"
#               define DKML_ABI_linux_arm32v7
#           endif /* __ARM_ARCH_6__ || ...,  __ARM_ARCH_7__ || ... */
#       elif __x86_64__
#           define DKML_ABI "linux_x86_64"
#           define DKML_ABI_linux_x86_64
#       elif __i386__
#           define DKML_ABI "linux_x86"
#           define DKML_ABI_linux_x86
#       elif defined(__ppc64__) || defined(__PPC64__)
#           define DKML_ABI "linux_ppc64"
#           define DKML_ABI_linux_ppc64
#       elif __s390x__
#           define DKML_ABI "linux_s390x"
#           define DKML_ABI_linux_s390x
#       endif /* __aarch64__, __arm__, __x86_64__, __i386__, __ppc64__ || __PPC64__,
__s390x__ */

Versioning and Contributing

Whenever a new ABI is added, it goes into a new version (ex. module V3). Your existing code that uses module V2 will be unaffected.

But each new ABI needs to have its own maintainer because I don't have access to every hardware platform on the planet!

For example, PowerPC (ppc64) and Linux on IBM Z (s390x) are supported in the C Header but not the OCaml module because there are no PowerPC and S390x maintainers.

Please consider contributing, especially if you want others to have an easier compilation story for your favorite hardware platform.

Full-Stack Web Dev in OCaml Tutorial w/ Dream, Bonsai, and GraphQL

Alexander (Sasha) Skvortsov announced

Hi everyone! I’ve written a tutorial blog series about full-stack web development in OCaml, and wanted to share it here.

Last semester, I took Penn State's CMPSC 431W, where our final project was to build a database-driven web application. Since I'm fairly familiar with web programming through my work on Flarum and past internships/side projects, I decided to use this opportunity to explore the OCaml web development ecosystem. I used Dream for the backend, and Bonsai for the frontend.

While working on this project, I realized two things:

  • OCaml is very underrated for web development. In addition to all the language’s great features and safety guarantees, the ecosystem is pretty good! Dream near-perfectly coincides with my vision of backend webdev, and Bonsai has a great balance of flexibility/elegance and safety.
  • I couldn’t find realistic but accessible full-stack web projects in OCaml available for reference. I found tutorials for bits and pieces, but nothing that connected all the dots.

I really enjoyed writing an article series on hardware design with OCaml, so I decided to do so for web development as well. In total, I wrote 7 articles that walk through my project’s:

  1. Full-Stack WebDev in OCaml Intro. This includes some background on the project, and instructions for accessing the live demo.
  2. Backend WebDev w/ Dream and Caqti.
  3. Building GraphQL APIs with Dream
  4. Setting up Bonsai.
  5. Understanding Bonsai. I actually wrote the first draft of this before I decided to do a blog, while trying to, well, understand Bonsai. It goes over some underlying concepts (SPAs, Frontend State Management, Algebraic Effects, Monads), as well as Bonsai’s core design.
  6. Using GraphQL in Bonsai.
  7. Routing in Bonsai and Project Conclusion.

Additionally, the project’s README has a comprehensive overview of the tech stack, folder structure, and usage instructions. It also includes some reflections on design decisions and my experience working with these libraries.

I had a lot of fun writing these, and I hope they’re useful to anyone considering OCaml for web development. Would be happy to answer any questions or comments.

Alexander (Sasha) Skvortsov later added

Also, forgot to mention this originally, but I recommend accessing the demo with one of the emails from this file or this file (all passwords are still here), as those users can also demo create/update functionalities.

Daniel Bünzli replied

People who are looking for more lightweight alternatives – and want to do web programming without bothering too much about front end insanity can have a look at hc (yes indeed: sending HTML over fetch, web programming excels at running in circles).

The front JavaScript for that CRUD webapp comes out at 132Ko uncompressed without even trying to tweak anything.

Sketch.sh now supports multiple compiler versions, starting with 4.13.1

Javier Chávarri announced

The interactive OCaml sketchbook sketch.sh has now support to store, edit and run sketches in different versions of the OCaml compiler.

Support for 4.13

Storing and running sketches using the compiler version 4.13.1 is now possible, this functionality has been added to the already existing support for version 4.06.1. The Reason parser and formatting tool refmt were also updated to a more recent version that supports 4.13.1.

Here you can see a sketch showcasing the monadic let syntax, using the example from the official OCaml docs: ZipSeq - Sketch.sh.

Existing sketches and forks

Previously existing sketches remain in 4.06.1, while newly created sketches will be on 4.13.1. For now, the only way to "migrate" a sketch to the newer version of the compiler is by copying its content and pasting it in a new sketch.

Forked sketches inherit the compiler version of the forked sketch.

Future plans

In the future, there are plans to support version 4.14.0 of the compiler, and we are considering adding a way so that the version of the compiler can be chosen for a given sketch. We are also working on migrating the editor UI codebase to a more recent version of ReasonReact, and use JSX3 instead of JSX2.

Feature requests and bugs

Please let us know in case you have a feature request, or if you encounter any issues or bugs. Also, don't hesitate to reach out via DM or any other means if you would like to contribute or participate in the project in some way.

Thanks to Ahrefs for supporting an Open Source Day initiative, which allowed to allocate time to work on this improvement for sketch.sh, and for providing the infrastructure to run the sketch.sh service for the community. Thanks as well to the authors and maintainers of the OCaml compiler, js_of_ocaml, and ReScript, that sketch.sh relies upon.

Explicit type binding and mutual recursion

Deep in this thread, octachron explained

For most use cases, if you want an explicit annotation for recursive function, it will be much simpler to use the type a. ... form:

let rec foo: type a. a -> a = fun x -> x
and bar: type a. a -> a = fun x -> foo x

This form is a shortcut for both adding an explicit universal quantified and and a corresponding locally abstract type (in other words ~let f : 'a . …. = fun (type a) -> … ~).

The root issue with

let rec f (type a) (x:a) = f x

is that the locally abstract type a is introduced after f. Moreover, without an explicit type annotation, a recursive function like f is monomorphic in its body and a monorphic function cannot be called on a type that was defined after the function.

In other words, the issue is that in term of type scopes, the function f is equivalent to

let f = ref Fun.id
type t = A
let x = !f A

which also fails with

Error: This expression has type t but an expression was expected of type 'a
       The type constructor t would escape its scope

This is why the second solution proposed by @Gopiandcode works. Indeed, in

let foo, bar = fun (type a) ->
  let rec foo (x: a) : a = x
  and bar (x: a) : a = foo x in
  foo, bar

the type a is defined before the recursive functions foo and bar, thus foo a does not break any scope constraint.


Gerd Stolpmann announced

findlib-1.9.4 is out. It mainly includes a change in the configuration script needed for OCaml-4-14.

For manual, download, manuals, etc. see here:


An updated OPAM package will follow soon.


Gerd Stolpmann announced

I just released omake-0.10.4, the build utility. This finally includes the fix for Apple Silicon, but also a couple of small changes (roughly everything since PR#100 to PR#146 on GitHub).

For docs and the download link see http://projects.camlcity.org/projects/omake.html. opam is underway.


If you happen to miss a CWN, you can send me a message and I'll mail it to you, or go take a look at the archive or the RSS feed of the archives.

If you also wish to receive it every week by mail, you may subscribe online.