OCaml Weekly News

Hello

Here is the latest OCaml Weekly News, for the week of September 03 to 10, 2= 024.

Oxidizing OCaml =E2=80=94 an update

Archive: https://discuss.ocaml.org/t/ann-oxidizing-ocaml-an-update/15237= /1

Diana Kalinichenko announced

Hi everyone! Last year, we made a series of blogposts describing our plans = to introduce Rust-like type system features to OCaml (see here= ). Now, we are sharing updates on everything we've done since last year for= ICFP 2024. Please read our blogpost and check out our compiler extensions at our Git= Hub!

Toy Autograd Engine in OCaml with Apple Accelerate Backend

Archive: https://discuss.ocaml.org/t/toy-au= tograd-engine-in-ocaml-with-apple-accelerate-backend/15239/1

John Jewell announced

I have been venturing to learn a new language and I landed on OCaml after h= earing a few interesting talks from Jane Street. I just made public a toy a= utograd engine in OCaml with an Apple Accelerate backend if anyone is inter= ested:=20

https://github.com/jew= elltaylor/camlgrad

I would really appreciate any feedback in terms of the OCaml code that I wr= ote so that I can improve. If anyone is willing to quickly take a look it w= ould mean a lot :slight_smile:

New release of cppo, with multi-line macros and higher-order m= acros

Archive: https://discuss.ocaml.= org/t/ann-new-release-of-cppo-with-multi-line-macros-and-higher-order-macro= s/15241/1

Fran=C3=A7ois Pottier announced

It is my pleasure to announce a new release of cppo (v1.7.0) with two new f= eatures.

The new syntax #def ... #enddef allows a macro definition to= span several lines, without backslashes. This syntax allows macro definiti= ons to be nested.
```
#def re=
peat_until(action,condition)
  action;
  while not (condition) do
    action
  done
#enddef
```

A parameterized macro can take a parameterized macro as a parameter: this i= s a higher-order macro.

#define=
 TWICE(e)=
          (e + e)
#define APPLY(F : [.], e) (let x=
 =3D (e) in F(x))
let forty_two =3D
  APPLY(<=
span style=3D"color: #242521; background-color: #fcf7ef;">TWICE,1+2+=
3+4+5+6)

For more details, please see the documen= tation.

OCamlPro's contributions to the 2024 ICFP in Milan

Archive: https://discuss.ocaml.org/t/ocamlpros-contr= ibutions-to-the-2024-icfp-in-milan/15244/1

OCamlPro announced

Today, a quick head's up about our contributions to this year's Internation= al Conference on Functional Programming which is unraveling right now in Milan!

This year, our team presents two topics:

"Snapshottable Stores": A generic and efficient data s= tructure for the implementation of backtracking algorithms, used particularly in automatic theorem provers and type checkers. This implementation in OCaml will soon be available on opam.
A presentation on opam, detailing the contents of the = latest major release 2.2, which was released in July, as well as how the op= am team operates.

Be sure to checkout the event, there are plenty of great pres= entations and video replays!

Until next time, which will be sooner than later with another one of our Fla= mbda2 Snippets,

Kind regards, The OCamlPro Team

Flambda2 Ep. 3: Speculative Inlining, by OCamlPro

Archive: https://discuss.ocaml.org/t/blog-flambda2= -ep-3-speculative-inlining-by-ocamlpro/15250/1

OCamlPro announced

As promised in our previous post about OCamlPro's = contributions to this year's International Conference of Functional Program= ming, we back again with a new entry in our Flambda2 Snippet blog se= ries!

F= lambda2 Ep. 3: Speculative Inlining covers inlining in general as well = as how our compiler handles it. We go in detail about how Speculative= Inlining allows more significant optimisations to take place.=20

This blog entry is key for a smooth read of our next article which will cov= er Upwards and Downwards Traversals in Flambda2.

Happy to say that it's already quite far down the release pipeline!

Frustrating Interactions with the OCaml Ecosystem while develo= ping a Synthesizer Library

Archive: https://discuss.ocaml.org/t/blog-frustrating-interactions-with-the-ocaml-e= cosystem-while-developing-a-synthesizer-library/15253/1

Steve Sherratt announced

https://www.gridbugs.or= g/frustrating-interactions-with-the-ocaml-ecosystem-while-developing-a-synt= hesizer-library/

Last year I made a synthesizer library in OCaml and had some struggles usin= g Dune and Opam, and also ran into several issues with libraries. I wrote a= blog post about all the problems I encountered.

Cmdlang - Yet Another CLI Library (well, not really)

Archive: https://discuss.ocaml.org/t/cmdlang-yet-anoth= er-cli-library-well-not-really/15258/1

Mathieu Barbin announced

I hope you had a nice summer! Mine took an unexpected turn when, roughly at= the same time, I wrote my first cmdliner subcommand and heard= about climate for the first time. My experience with OCaml CL= I so far had been centered around core.command.

When I read climate's terminology section and how it defines Terms, Arguments, and Parameters, someth= ing clicked. Seeing how climate's API managed to make position= al and named arguments fit so nicely together, I thought: "Wow, for the fir= st time, it seems I'll be able to write a CLI spec on a whiteboard without = referring to some code I never seem to get right (I am looking at you, core.command's anonymous arguments)."

I got quite excited and thought: "Can I switch to climate toda= y?" But reality checked: it's not on opam yet, still under construction, I'= m not sure what the community will do, etc.

Implementing my own engine for an API resembling climate felt = like a wasted effort, knowing about the work happening in climate. Still, having a 'a Param.t, 'a Arg.t, and 'a Command.t type that I would get to know and love felt too goo= d to pass up.

I stared at the climate types for a while, and filled with hap= py thoughts about a bright CLI future, it occurred to me: can I use an API = like climate but compile it down to existing libraries such as= cmdliner or core.command? (and climate too!). I wrote down the following types:




Climate


'a Param.t     -> 'a Climate.Arg_parser.=
conv
'a Ast.Arg.t   -> 'a Climate.Arg_parser.t
'a Command.t   -> 'a Climate.Command.t






Cmdliner


'a Param.t     -> 'a Cmdliner.Arg.conv
'a Arg.t       -> 'a Cmdliner.Term.t
'a Command.t   -> 'a Cmdliner.Cmd.t


core.command



'a Param.t     -> 'a core.Command.Arg_type.t
'a Arg.t       -> 'a core.Command.Param.t
unit Command.t -> core.Command.t




… which I interpreted as stating the following theorem:




There exists an abstraction to encode OCaml CLIs that lives in the intersec=
tion of what's expressible in other well established libraries.




"One EDSL to command them all," so to speak. I couldn't resist the temptati=
on to build actual terms for these types. That gave birth to cmdlang.

As a test, I switched one of my projects to cmdlang, with cmdliner as a backend. I liked the changes I made in the process. The 1-line bin/main.ml is now the only place that specifies which backend I want to use; the re= st of the code is programmed solely against the cmdlang API. T= his means I'll be able to easily experiment with compiling down to cl= imate in the future.

I am not against the multiplicity of solutions in general, but I tend to fe= el uneasy when incompatible libraries emerge, partitioning the ecosystem. A= s a community, we know too many examples of this. In this instance, I want = to call the core.command vs cmdliner situation a = … cli-vage.

I don't see my work on cmdlang as competing with these other l= ibraries. Quite the contrary, it makes it easier for me to experiment with = them without much changes while exploring the subject of CLI in general. Al= so, as a library author, if you wish to expose CLI helpers to your users, a= library like cmdlang will give you a pleasant way to do so, a= s you can express your helpers with it, knowing your users will have the ch= oice to translate them to the backend of their choice.

Before writing this post, I had a very pleasant chat with @gridbugs. I want= to make it clear that I do not think cmdlang is competing wit= h climate either. I think climate is a very promi= sing library and I believe it will, in due time, deliver auto-completion to= many - this has been a highly anticipated feature within the community. I = wish to dedicate the initial work that I did on cmdlang to @gr= idbugs due to the impactful influence climate had on my work, and how it he= lped me improve my general understanding of declarative CLI libraries.

These are very early days for cmdlang. There are still areas I= am fuzzy on, and I haven't really validated the whole design yet. I have p= ut some thoughts in this Future Plans page. One thing that I did not = initially include there would be to explore the feasibility of writing a mi= ni-compiler for cmdlang targeting stdlib.arg as a= runner. I am not sure how much you'd end up restricting cmdlang for this to work. I thought that'd be a fun project to tackle at a futur= e point, as it would make a nice addition to the overall architecture of th= e project.




I'd welcome your input to help me shape the future of cmdlang =
if you have an interest in this project.



Thanks for reading!


zarr v0.1.0


Archive: ht=
tps://discuss.ocaml.org/t/ann-zarr-v0-1-0/15259/1


zoj613 announced


Hi everyone, I'd like to announce the first release of zarr, a=
n Ocaml implementation of the Zarr version 3 storage format specification for chunked & compressed multi-dimensional arrays, designed for use i=
n parallel computing.










why?


The project was mainly inspired by the lack of functional programming langu=
age implementations of this specification as shown in this implementations table. Since I have been =
learning OCaml these past few months I figured I'd take on the challenge of=
 producing the first functional programming implementation of Zarr, and it =
was a great learning experience!


Features


Supports creating n-dimensional Zarr arrays and chunking them along any=
 dimension.
Compress array chunks using a variety of supported compression codecs.<=
/li>
Supports indexing operations to read/write views of a Zarr array.
Supports storing arrays in-memory or the local filesystem. It is also
extensible, allowing users to easily create and use their own custom storag=
e backends. See the example implementing a Zip file store fo=
r more details.

Supports both synchronous and concurrent I/O via Lwt and <= code>Eio.
Leverages the strong type system of Ocaml to create a type-safe API; ma= king it impossible to create, read or write malformed arrays.
Supports organizing arrays into hierarchies using Groups.

Example

Below is a demo of the library's API for creating, reading and writing to a= Zarr hierarchy.

open Zarr
open Zarr.Metadata
open Zarr.Node
open Zarr.Codecs
open Zarr.Indexing
open Zarr_sync.Storage
(* opens infix operators >>=3D =
and >>| for monadic bind & map *)
open FilesytemStore.Deferred.Infix

let store =3D Files=
ystemStore.create "testdata.zarr" in
let group_node =3D =
GroupNode.of_path "/some/group" in
FilesystemStore.create_group store g=
roup_node;
let array_node =3D =
ArrayNode.(group_node / "name");;
(* creates an array with char data ty=
pe and fill value '?' *)
FilesystemStore.create_array
  ~codecs:[`Transpose [|2; 0; 1|]; `Bytes BE; `Gzip L2]
  ~shape:[|100; 100; 50|]
  ~chunks:[|10; 15; 20|]
  Ndarray.Char=20
  '?'
  array_node
  store;
let slice =3D [|R [|0; 20|]; I 10; R [||]|] in
let x =3D Filesyste=
mStore.read_array store array_node slice Ndarray.Char in
(* Do some computation on the array s=
lice *)
let x' =3D Zarr.Nda=
rray.map (fun _ -> Random.int 256 |> =
Char.chr) x in
FilesystemStore.write_array store ar=
ray_node slice x';
let y =3D Filesyste=
mStore.read_array store array_node slice Ndarray.Char in
assert (Ndarray.equal x' y);

Installation

The library comes in several flavors depending on the synchronous / asynchr= onous backend of choice. To install the synchronous API, use

$ opam install zarr-sync

To install zarr with an asynchronous API powered by Lwt or Eio, use

$ opam install zarr-lwt
$ opam install zarr-eio

The documentation can be found here and the source code there

I'm happy to answer any questions regarding the library and more than welco= me suggestions for improvements (especially performance!), issue reports as= well as PR's.

Brr 0.0.7

Archive: http= s://discuss.ocaml.org/t/ann-brr-0-0-7/15263/1

Daniel B=C3=BCnzli announced

There'a new release of Brr, an ISC licenced toolkit for programming browsers with the js_of_oc= aml compiler.

This release has some changes to support work being done for wasm_of_= ocaml; thanks to @vouillon for his patches. There are also other sma= ll fixes and additions, consult the release notes for the details and thanks to all the contribut= ors.

A big thanks to my donator= s for supporting my work. I welcome the (not so[^1]) new donator Tarides.

Home page, Docs and manuals or odig doc= brr

Install: opam install brr (PR)

Best,

Daniel

[^1]: Tarides has been generously donating for my work from the onse= t but used to do it via the Mirage organisation.

Ocsigen Server 6.0.0

Archive: https://discuss.ocaml.org/t/ann-ocsigen-server-6-0-0/15265/1

Vincent Balat announced

We're delighted to announce a major new version of Ocsigen Server! This ver= sion 6.0.0 focuses on the use of Ocsigen Server as a library, without any c= onfiguration file, which is now much easier and brings the exact same featu= res as the executable.

Example of use:

To add a Web server to your OCaml program, serving static files from direct= ory `static`:

let _ =3D Ocsigen_server.start=20
          [ Ocsigen_server.host [Staticmod.run ~dir:"static" ()]]

Install:

opam install ocsigenserver

Example of Dune file for this program:

(executable
 (public_name myproject)
 (name main)
 (libraries
  main
  ocsigenserver
  ocsigenserver.ext.staticmod))

Compile with:

dune build

Ocsigen Server can of course still be used as an executable taking its conf= iguration from a file. This allows for non OCaml developers to use it and m= ake their own configurations. It also makes it possible to distribute a bin= ary version of your Web applications.

Ocsigen Server is build in modular and extensible way. The default opam pac= kages comes with several extensions. In the example above, we are using Sta= ticmod for serving static files. Other extensions makes it possible for exa= mple to configure redirections, to control the access to some sub-directory= , to use a reverse proxy, to rewrite the request, compress the output etc.

The programming interface follows exactly the structure of the configuratio= n file: instructions are tried in order until one generates a result, then = some other instructions can be used to change the result (like compressing = it or adding some headers).

Here is a more complex example of configuration:

let _ =3D
  Ocsigen_server.start
    ~ports:[`All, 8080]
    ~command_pipe:"local/var/run/mysite-cmd"
    ~logdir:"local/var/log/mysite"
    ~datadir:"local/var/data/mysite"
    ~default_charset:(Some "utf-8")
    [ Ocsigen_server.host
        ~regexp:"mydomain.com"
        [ Ocsigen_server.site ["subsite"]
            [ Accesscontrol.(
                if_
                  (and_
                     [ ip "122.122.122.122"=

                     ; header ~name:=
"user-agent" ~regexp:".*FooBar.*"
                     ; method_ `POST ])
                  [forbidden] [])
            ; Authbasic.run ~realm:"myre=
alm"
                ~auth:(fun _u p -> Lwt.return (p =3D "toto"))
                ()
            ; Staticmod.run ~dir:"local/=
var/www/otherdir" () ]
        ; Ocsigen_server.site ["othersubsite"]
            [ Revproxy.run
                ~redirection:
                  (Revproxy.create_r=
edirection ~full_url:false ~regexp:"(.*)"
                     ~keephost:true "htt=
p://localhost:8888/\\1")
                () ]
        ; Redirectmod.run
            ~redirection:
              (Redirectmod.create_re=
direction ~full_url:false ~regexp=
:"old(.*)"
                 "new\\1")
            ()
        ; Staticmod.run ~dir:"local/var=
/www/staticdir" ()
        ; Cors.run ~max_age:86400 ~crede=
ntials:true ~methods:[`POST; `GET; `HEAD]
            ~exposed_headers:
              [ "x-eliom-application"
              ; "x-eliom-location"
              ; "x-eliom-set-process-cookie=
s"
              ; "x-eliom-set-cookie-substit=
utes" ]
            ()
        ; Deflatemod.run
            ~mode:
              (`=
Only
                [ =
Some "text", Some "html")
                ; =
Some "text", Some "javascript")
                ; =
Some "text", Some "css")
                ; =
Some "application", Some "javascript")
                ; =
Some "application", Some "x-javascript")
                ; =
Some "application", Some "xhtml+xml")
                ; =
Some "image", Some "svg+xml")
                ; =
Some "application", Some "x-eliom") ])
            () ] ]

In this example, the server defines one virtual host for domain mydom= ain.com. It will first check whether it is a request for directory <= code>subsite/, and if yes, will reject the request with 403 Forbidde= n if it is a POST request coming from user-agent FooBar at IP = 122.122.122.122. If not, it will ask for a password before serving files fr= om directory local/var/www/otherdir. Then we define another subsite othersubsite for which the requ= ests will be transfered to another Web server running locally on port 8888,= then rewrite the answer location header accordingly. Then, if the page is still not generated, the server will send a redirectio= n if URLs starts with "old". Otherwise, it will try to serve files from directory local/var/www/st= aticdir. If the page has still not been found, a 404 Not found will be = sent, otherwise, some CORS headers will be added, and the result will be co= mpressed before being sent.

Compile this example with the following dune file:

(executable
 (public_name myserver)
 (name main)
 (libraries
  ocsigenserver
  ocsigenserver.ext.staticmod
  ocsigenserver.ext.authbasic
  ocsigenserver.ext.extendconfiguration
  ocsigenserver.ext.outputfilter
  ocsigenserver.ext.cors
  ocsigenserver.ext.accesscontrol
  ocsigenserver.ext.deflatemod
  ocsigenserver.ext.redirectmod
  ocsigenserver.ext.revproxy
 ))

Eliom and Ocsigen Start have also been updated for being used without confi= guration file and will be released very soon.

dream-html and pure-html

Archive: https://discuss.ocaml.org/t/ann-dream-html-pure-html-3-5-2/148= 08/4

Yawar Amin announced

[ANN] dream-html & pure-html 3.6.1, 3.6.2

A double announcement:

3.6.1: when in XML rendering mode, correctly render empty-value attributes = as having an empty string value. Thanks to @jonsterling !

3.6.2: automatically switch to XML rendering mode when rendering SVG and Ma= thML tags inside HTML rendering mode.

Advanced Code Navigation coming to OCaml-LSP

Archive: https://discuss.ocaml.org/t/ann-advanced-cod= e-navigation-coming-to-ocaml-lsp/15266/1

PizieDust announced

Jump to Target

Currently, the standard LSP protocol only allows for generalized code navig= ation (`goto definition`, `goto declaration`, `goto implementation`, `goto = type-definition`), which is not very useful when it comes to precise moveme= nts.

Coming to OCaml-lsp soon will be the ability to jump from one point in your= code to another based on Merlin's Ju= mp command.

Implementing this functionality took a bit of thinking as we wanted a solut= ion that works for all supported editors (Vscode, Emacs and Vim) without an= y additional specific client implementations. We used a combination of call actions plus the LSP showDocumentRequest to = move the cursor to the interesting position.

The call actions display are contextual and will display only if it's relev= ant to the code under the cursor.

Here is a demo in VSCode.

3D"618bfd77db1=

Other OCaml News

From the ocaml.org blog

Here are links from many OCaml blogs aggregated at the ocaml.org blog.

Old CWN

If you happen to miss a CWN, you can send me a message and I'll mail it to you, or go take a loo= k at the archive or the <= a href=3D"https://alan.petitepomme.net/cwn/cwn.rss">RSS feed of the archive= s.

If you also wish to receive it every week by mail, you may subscribe to the= caml-list.

Alan Schmitt

OCaml Weekly News

Table of Contents

Oxidizing OCaml =E2=80=94 an update

Diana Kalinichenko announced

Toy Autograd Engine in OCaml with Apple Accelerate Backend

John Jewell announced

New release of cppo, with multi-line macros and higher-order m= acros

Fran=C3=A7ois Pottier announced

OCamlPro's contributions to the 2024 ICFP in Milan

OCamlPro announced

Flambda2 Ep. 3: Speculative Inlining, by OCamlPro

OCamlPro announced

Frustrating Interactions with the OCaml Ecosystem while develo= ping a Synthesizer Library

Steve Sherratt announced

Cmdlang - Yet Another CLI Library (well, not really)

Mathieu Barbin announced

Climate

Cmdliner

core.command

zarr v0.1.0

zoj613 announced

why?

Features

Example

Installation

Brr 0.0.7

Daniel B=C3=BCnzli announced

Ocsigen Server 6.0.0

Vincent Balat announced

dream-html and pure-html

Yawar Amin announced

Advanced Code Navigation coming to OCaml-LSP

PizieDust announced

Jump to Target

Other OCaml News

From the ocaml.org blog

Old CWN