OCaml Weekly News

The Dune team is pleased to announce the release of dune 3.24.2.

See the full changelog for the new fixes, and for attribution to the contributors who made it all possible. Thank you, contributors!

If you encounter a problem with this release, please report it in our issue tracker.

I am glad to announce Merlin 5.8.1-505, a new release that notably fixes a caching issue that manifested frequently when using OCaml-LSP with Dune in watch mode.

We also released preview versions of both Merlin and OCaml-LSP working with today's OCaml compiler's trunk. You can use them to hack on the compiler, but they will break again with futures changes to the types AST.

Hello!

it's my pleasure to announce a (very) preliminary version of OCaml-sdl3 (or simply sdl3 for the package name), which obviously is a set of bindings for the SDL3 library.

https://github.com/sanette/ocaml-sdl3

Why announcing a preliminary version? First, because I won't have much time to work on it in the next couple of months, and second, because I believe you can already have fun with it!

The bindings are generated by a program that uses clang to inspect the SDL code. This is a first phase; not all functions are usable yet. I have started to write manual wrappers for a few of them. (See the examples).

AI disclaimer: Although I have to admit chatGPT was my friend for understanding how to use clang and ctypes, I would like to state here that the whole binding generator (plus all manual additions, of course) was entirely written by (my) hand.

This message is intentionally written in French. It is a call for papers for the “Francophone Days on Functional Languages” to be held at the end of January 2027 in Britany. Papers can be written in English, but the presentations themselves are expected to be given in French.

Merci de faire circuler : premier appel à communications

JFLA 2027 : Journées Francophones des Langages Applicatifs

http://jfla.eu/jfla2027.html

26 janvier au 29 janvier 2027

Maison Saint-François

Les 38èmes Journées Francophones des Langages Applicatifs (JFLA) se tiendront en Bretagne, à Dinard (Ille-et-Vilaine), du mardi 26 janvier 2027 au vendredi 29 janvier 2027.

Les JFLA réunissent concepteur·rices, utilisateur·rices et théoricien·nes ; elles ont pour ambition de couvrir les domaines des langages applicatifs, de la preuve formelle, de la vérification de programmes, et des objets mathématiques qui sous-tendent ces outils. Ces domaines doivent être pris au sens large : nous souhaitons promouvoir les ponts entre les différentes thématiques.

• Langages fonctionnels et applicatifs : sémantique, compilation, optimisation, typage, extensions à d'autres paradigmes.
• Assistants de preuve : implémentation, nouvelles tactiques, développements présentant un intérêt théorique, technique ou méthodologique.
• Logique, correspondance preuve-programme, réalisabilité, extraction de programmes, modèles.
• Spécification, prototypage, développements formels d'algorithmes.
• Vérification de programmes ou de modèles, vérification déductive, interprétation abstraite, raffinement.
• Utilisation industrielle des langages fonctionnels et applicatifs, ou des méthodes issues de la communauté scientifique. Outils et plateformes pour le web.
• Enseignement ou diffusion des langages fonctionnels et applicatifs. Environnements et méthodes de développement, retours d'expérience.

Les articles soumis aux JFLA sont relus par au moins deux personnes s'ils sont acceptés, et au moins trois personnes s'ils sont rejetés. Les critiques du comité de programme sont toujours bienveillantes et la plupart du temps encourageantes et constructives, même en cas de rejet.

Il n'y a donc pas de raison de ne pas soumettre aux JFLA !

DATES IMPORTANTES

/!\\ Attention : les dates limites sont fermes et définitives. Il n'y aura pas d'extension. /!

• Soumission des résumés et articles : 16 octobre 2026, GMT+2
• Notification aux auteurs et autrices : 2 décembre 2026, GMT+2
• Version finale des articles : 16 décembre 2026, GMT+2

SOUMISSIONS

Nous acceptons deux types de soumissions :

• Article de recherche (18 pages max.) portant sur des travaux originaux. Nous acceptons des travaux en cours, pour lesquels l'aspect recherche n'est pas entièrement finalisé. Nous encourageons aussi la soumission d'articles présentant avec élégance un résultat connu sous un angle nouveau.
• Article court (9 pages max.) décrivant un problème particulier, les pistes en cours d'investigation, et visant à rechercher de l'aide de la part de la communauté. Les articles courts peuvent également présenter de manière synthétique et cohérente des résultats déjà publiés. Enfin, ils peuvent présenter un outil logiciel dont l'exposé constituera une démonstration.

CONSIGNES AUX AUTEURS ET AUTRICES

Les articles peuvent être rédigés en français ou en anglais.

La forme de l'article doit être soignée, et le contenu rédigé de manière structurée et claire.

Le style LaTeX jflart doit impérativement être utilisé sans modification de la mise en page. Le style LaTeX et sa documentation sont disponibles depuis le site web de la conférence.

Les limites de pages sont strictes. Les références bibliographiques ne sont pas comptabilisées dans la limite de pages. Les annexes aux articles ne sont pas autorisées.

Les auteurs et autrices peuvent soumettre du matériel supplémentaire, séparé de l'article soumis, sous forme de texte (version longue, sans limite de pages) et/ou de développement logiciel. L'évaluation de ce matériel supplémentaire est à la discrétion du comité de programme. Les articles soumis doivent donc être auto-contenus et évaluables sans ce matériel supplémentaire.

Les soumissions parallèles dans d'autres conférences, journaux ou workshops avec actes ne sont pas autorisées.

Les membres du comité de programme sont autorisés à soumettre un article. Les présidentes du comité ne le sont pas.

Les articles doivent être soumis via le site :

https://types-hotcrp.paris.inria.fr/jfla27

L'évaluation des articles suit un processus en simple-aveugle : les rapports des articles sont anonymes, mais pas les auteurs et autrices.

Les articles acceptés seront publiés dans les actes de la conférence, sur HAL, et les auteurs et autrices en donneront une présentation lors des journées. Les présentations seront, de préférence, données en français.

Yannick ZAKOWSKI et Stefania DUMBRAVA

JFLA 2027

Hello,

After some feedback we've received about Hegel, I am happy to announce v0.14.0 has removed the dependency on Core. The Core generators from the previous versions now exist in an optional sublibrary called Hegel_jane. Thanks to @c-cube for his help with his release!

For slightly more detail, see the release notes:

https://github.com/hegeldev/hegel-ocaml/releases#release-v0.14.0

Hello everyone,

ifs-fractals — a small library and CLI that draws the attractors of iterated function systems (the Barnsley fern and friends) with the chaos game — just got its first feature release since arriving on opam this summer. After sixteen years of intensive development ;-)

What's new in 1.1.0:

• Headless PNG rendering. Until now the fractals could only be drawn into a graphics window. You can now render straight to a file, no display needed:

  ifs-fractals -o fern.png -s 800x1280 barnsley
  

• or from code, save_png barnsley 200000 "fern.png". This makes the package usable on servers and in CI, and made testing much easier. A fun constraint I set myself: no new dependencies — the PNG writer is self-contained, writing 8-bit PNGs with stored deflate blocks in pure OCaml.

• Density coloring. The chaos game doesn't visit the attractor uniformly — some regions are hit far more often than others, and that density is where much of the visual structure lives. With -c (or save_png \~color:true), pixels are colored by their hit count on a log scale. The image above is the result: the fern gets its veins back.

  ifs-fractals -c -o fern.png -s 800x1280 barnsley
  

Install or upgrade with opam install ifs-fractals, then try ifs-fractals –list to see all sixteen attractors.

If you're curious how new attractors get designed — how one finds the six coefficients per affine map that produce a maple leaf or a sunflower — I wrote the techniques up in a short paper, Building New IFS Attractors: a Working Vocabulary of Affine Maps ( https://doi.org/10.5281/zenodo.21804778 ), with a more narrative account on my blog ( https://www.cedricbonhomme.org/2026/07/27/challenging-claude-with-ifs-fractals/ ).

One disclosure, since it's the interesting part of the story: the new attractors were designed by Claude (Anthropic's Fable 5 model) — the blog post recounts that experiment. Everything else — the library, the CLI, the PNG writer, and the paper itself — is my own work, not LLM output.

• Repository: https://github.com/cedricbonhomme/iterated-function-systems
• opam: https://opam.ocaml.org/packages/ifs-fractals/
• Changelog: https://github.com/cedricbonhomme/iterated-function-systems/blob/master/CHANGES.md

Feedback and new fractal definitions are very welcome. Just make a pull-request.

If you design an attractor you like, I'd love to see it!

A new security advisory is available. As usual, it can be found at https://github.com/ocaml/security-advisories and https://osv.dev/list?q=&ecosystem=opam

Best,

Hannes

id: OSEC-2026-14
modified: "2026-08-07T13:00:00Z"
published: "2026-08-07T13:00:00Z"
severity: "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:L"
severity_score: "4.3 (Medium)"
affected: "mirage-crypto-pk" {< "2.3.0"}
events: [
   [
     git "https://github.com/mirage/mirage-crypto.git" [
       [fixed "a0f59a0c90eb067505b55a03d3bb104eacd6dd33"]
     ]
   ]
]
credits: [
   [reporter "Thomas Gazagnaire"]
   [remediation_developer "Hannes Mehnert"]
]
cwe: [ CWE-248 ]
affected_bindings: [
   "Mirage_crypto_pk.Rsa.encrypt"
   "Mirage_crypto_pk.Rsa.decrypt"
   "Mirage_crypto_pk.Rsa.PKCS1.encrypt"
   "Mirage_crypto_pk.Rsa.PKCS1.decrypt"
   "Mirage_crypto_pk.Rsa.PKCS1.sig_encode"
   "Mirage_crypto_pk.Rsa.PKCS1.sig_decode"
   "Mirage_crypto_pk.Rsa.PKCS1.sign"
   "Mirage_crypto_pk.Rsa.PKCS1.verify"
   "Mirage_crypto_pk.Rsa.OAEP.encrypt"
   "Mirage_crypto_pk.Rsa.OAEP.decrypt"
   "Mirage_crypto_pk.Rsa.PSS.sign"
   "Mirage_crypto_pk.Rsa.PSS.verify"
]

A new security advisory is available. As usual, it can be found at https://github.com/ocaml/security-advisories and https://osv.dev/list?q=&ecosystem=opam

Best,

Hannes

id: OSEC-2026-15
modified: "2026-08-07T13:00:00Z"
published: "2026-08-07T13:00:00Z"
severity: "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:L"
severity_score: "4.3 (Medium)"
affected: "mirage-crypto-ec" {< "2.3.0"}
events: [
   [
     git "https://github.com/mirage/mirage-crypto.git" [
       [fixed "1f0bf67044e67cf6e46911fcd77a0ff706b6c3e7"]
     ]
   ]
]
credits: [
   [reporter "Thomas Gazagnaire"]
   [remediation_developer "Hannes Mehnert"]
]
cwe: [ CWE-125 CWE-248 ]
affected_bindings: [
   "Mirage_crypto_ec.P256.Dsa.pub_of_octets"
   "Mirage_crypto_ec.P256.Dh.key_exchange"
   "Mirage_crypto_ec.P384.Dsa.pub_of_octets"
   "Mirage_crypto_ec.P384.Dh.key_exchange"
   "Mirage_crypto_ec.P521.Dsa.pub_of_octets"
   "Mirage_crypto_ec.P521.Dh.key_exchange"
]

Hi :​) Wanted to announce a new package I worked on over the past week I just published, floatml! It's a tiny wrapper around Rust's soft float library (rustc_apfloat), enabling fast and efficient IEEE-754 float16, float32, float64 and float128 in OCaml. I also made sure the code allocates as little as possible, so it should be reasonably efficient.

Disclaimer: a good chunk of the code is LLM-authored. I chose the interface I wanted and guided it, but the glue code with Rust is LLM generated (and extensively tested!)

I had to create it for Soteria, where we need reductions for concrete float operations, to avoid using Z3 when possible. Maybe you'll find a use for it too!

Hi all!

I have just pushed a full reboot of the Android cross-compiler opam repository!

The reboot follows the most recent practices set by the windows counter part and is pretty clean, thanks to the vast improvements made in cross-compiling support in the main OCaml compiler recently.

Not all packages have been migrated, due to lack of time. The old branch is still available at old-main.

This is a community repository so, if cross-compiling OCaml code is in your wheelhouse, feel free to submit some patches!

Hi all!

I am happy to announce that Ravel 1.0.1 is now available on opam.

Ravel is a small functional language that compiles to interaction nets — a graph-rewriting model with strong confluence and implicit parallelism. It features user-defined algebraic data types, deep pattern matching, first-class functions with lexical closures, and automatic resource management via labeled duplication and erasure nodes.

https://opam.ocaml.org/packages/ravel/

Feedback and contributions are welcome!

Hi!

We are happy to announce the first public release of p4spectec, a framework for mechanizing the P4 language specification. It provides a domain-specific language for writing formal specifications in the form of algorithmic inference rules (Lee et al., 2026).

A specification written in P4-SpecTec is executable. By writing typing rules with P4-SpecTec, you get a reference type checker. By writing dynamic semantics rules, you get a reference interpreter. The prose backend also generates human-readable documentation from the mechanized specification.

https://opam.ocaml.org/packages/p4spectec/

For detailed instruction, source code and the mechanized P4 specification, check out the GitHub repository linked to the package.

I'm happy to announce that first public release of Neo4j Driver.

It is a pure Ocaml driver for Neo4j database build for Eio. It is intended to reach full feature parity with official drivers. Although it's not there yet, at this stage it's fully usable with the following features:

• Bolt protocol 3.0, 4.2–4.4, 5.0–5.8 and 6.0.
• Plain bolt:// and TLS bolt+s:// / bolt+ssc:// connections.
• Auto-commit queries with lazy streaming results.
• Both explicit and managed transactions with automatic retry.
• Bookmarks.
• Temporal types with named time zones (embedded IANA database plus an LMT fallback before 1970).
• Basic authentication (LOGON after HELLO on Bolt >= 5.1).
• TestKit conformance: 114 of 126 tests passing (12 skipped).

https://opam.ocaml.org/packages/neodriver/

Dear everyone,

I'm happy to announce that opam-audit has been released to opam in its initial version. It is an opam plugin (so you install it in one switch and have it available in all switches as opam audit).

So, opam-audit keeps a local copy of the security-advisory database from the OCaml Security Team. When you run opam audit, it will inspect all installed packages in your switch against the known advisories and report which installed packages have a known security vulnerability.

It either returns 0 if no known vulnerable packages were found, or non-zero and a list of installed vulnerable packages. This way, your CI systems can before publishing your application call opam audit and ensure that no vulnerable package sneaked into your binary.

An earlier mention in here https://discuss.ocaml.org/t/request-for-comments-what-to-do-with-opam-packages-that-have-known-security-vulnerabilities/18087/2

Source code at https://github.com/hannesm/opam-audit - bug reports and feature requests are very welcome.

Dear all,

I’m happy to announce the release shakuhachi v0.3.2, which also contains 0.3.0 and 0.3.1 that I forgot to announce.

Shakuhachi is still a music collection manager.

The TL;DR of those releases:

• Regex substitution is slow and replacing it vastly improves the speed of formatting.
• Fixing quite some wrong behaviors and my own mistakes.
• A lot of new functions exposed which make it a somewhat usable foundation to built things on top.
• Fighting against OCaml's linkage strategy to allow plugins (using dynlink) to have all the needed dependencies.
• Few new options (eg. sorting items, avoiding inserting duplicate songs, …).

To see the list of all changes, you can read the changelog.

https://codeberg.org/EruEri/shakuhachi

Thanks you for your time

Sincerely yours.

Yo!

I am happy to announce the first public release of Bytream, a library for streaming bytes and crunching them.

This library contains angostic I/O runtime mechanisms for organization byte streams processing for write and read in an efficient manner, e.g. for buildings codecs and protocol implementations.

The idea and motivation for creating this project arose from the need to effectively parse all kinds of binary formats and protocols. Existing tandem Angstrom and Faraday is great, but it is more expressive than I need. And Bytesrw and Iostream talk about how to receive and send data, not how to process or record it.

That's why Bytream solves this problem. It borrows ideas from slices and lifetimes from ByteSw and unbuffering from Angstrom, allowing you to write regular OCaml code like with the Stdlib channels, which work in streaming mode at any I/O runtime.

Smaji Glyph Path

An OCaml library for manipulation of font glyph outlines. It provides a suite of tools for analyzing, transforming, composing, and converting glyph paths between different formats.

This library supports two primary path description standards:

1. SVG Path: Supports <path> element (including its d attribute) and optional <g> element as containers.
2. Glif Format: Supports the data used to describe glyph outlines in the Unified Font Object (UFO) format.

It has published on opam, so it can be installed easily by opam install smaji_glyph_path

And here are two use case samples

Smaji Glyph Path

Smaji Glyph Outline Description

Smaji Glyph Stroke Description

I’d like to introduce two practical applications of the Smaji Glyph Path library: Smaji Glyph Outline Description(god) and Smaji Glyph stroke description(gsd). Currently, these aren't published on opam yet; I've just provided the links here as examples of how you can use the Smaji Glyph Path library. If you find these examples helpful, please let me know by replying or giving it a "like"—that'll let me know there's interest so I can get them onto opam for better maintenance.

I’ve designed a markup language that makes it very easy to build glyphs by combining different strokes. Thanks to Smaji Glyph Path, creating a glyph is basically like playing with building blocks—you can even generate an animation of the entire writing process!

Check out these examples:

Example 1: Building the "八" character by combining two strokes.

<god version="1.0">
  <glyph unicode="516b,0">
    <stroke type="t" x="0" y="0" width="56" height="112"/>
    <stroke type="p" x="76" y="0" width="56" height="112"/>
  </glyph>
</god>

Example 2: Combining an existing "god" (不) and a horizontal stroke (一) to create "丕" (with animation).

<god version="1.0">
  <glyph unicode="4e15,0">
    <character utf8= "不" x="0" y="0" width="128" height="120"/>
    <stroke type="h" x="0" y="114" width="128" height="14"/>
  </glyph>
</god>

If this image is still, try to reload the image or open this image in a new tab or window to restart the animation.

After using GOD for a while, I realized it’s actually better suited for phonetic scripts rather than ideograph. Because of that, I designed a new library specifically for ideograph (especially Chinese): Smaji Glyph Stroke Description (GSD).

For a deeper dive into GOD, the differences between it and GSD, and the pros/cons of each use case, check out the links at the beginning. I'll keep it brief here—feel free to check those out if you're interested!