OCaml Weekly News

We are pleased to announce the release of ocamlformat 0.14.2. This minor release improves the recent 0.14.0 and 0.14.1 releases regarding the doc-comments option.

ICFP, and by extension the ML workshop, will be now officially be held online with a significantly reduced fee. Due to the change in official status we decided to extend the submission deadline to the end of May.

Dear OCamlers, I am happy to pre-announce memprof-limits, an implementation of per-thread global memory limits, and per-thread allocation limits à la Haskell, compatible with systhreads.

Memprof-limits interrupts the execution by raising an asynchronous exception, an exception that can arise at almost any location in the code. I also announce a guide on how to recover from asynchronous exceptions and other unexpected exceptions that you find in the documentation. It summarises knowledge acquired in OCaml by the Coq proof assistant as well as in other programming languages. To my knowledge, this has never been told in OCaml textbooks, so I thought it might be of general interest to you. This research is part of a wider work aiming to regulate the use of asynchronous exceptions in OCaml in coordination with multicore language designers.

Global memory limits let you bound the memory consumption inside specific parts of your program, in terms of memory used by the whole program. It is inspired by this other post, but in a form readily available for use with systhreads.

Allocation limits let you bound the execution of parts of the program measured in number of allocations, analogous to the same feature in Haskell advocated in a nice post by Simon Marlow. Allocation limits count allocations but not deallocations, and is therefore a measure of the work done, which can be more suitable than execution time.

Memprof-limits, as the name tells, uses the upcoming Memprof engine from OCaml 4.11, with a low sampling rate that does not affect performance. A reimplementation of the Memprof interface compatible with memprof-limits running at the same time is provided for profiling needs.

Memprof-limits is available on the public opam repository, but depends on OCaml 4.11 which at the moment is available from the beta opam repository only. It is experimental for reasons explained in the manual.

Tezos executables and libraries have just been released on opam. You can thus build them from source with a simple opam install tezos and build your own projects upon them.

Tezos has a soft-updating mechanism that works (roughly) as follows:

The network starts with a genesis protocol (“protocol” here means “economic protocol”: the rules according to which smart contracts are initiated and acted upon, transactions take place, etc.) in which a single public key is specified.

The genesis protocol has no notion of coin, currency, smart-contract, etc. Instead, the genesis protocol knows a single operation: a protocol injection.

The protocol injection for genesis requires the operation to be signed by the private key that matches the public key of the genesis block. And the protocol injection changes, irreversibly, the genesis protocol to a new protocol. This new protocol specifies what constitutes a valid block to add to the chain.

In the Tezos blockchain, the protocol injected on top of genesis included a notion of coins and an in-protocol voting system to inject new protocols based on consensus amongst coin-holders. There is even a system to obtain the protocol sources over the blockchain network so they can be compiled by each node and dynlinked directly in: you don't need to update/restart your node to get the protocol updates. However, this is arbitrary: you can start a new block-chain with a different protocol.

For example, you could re-implement Bitcoin (proof-of-work, coins+transfer, etc.) as a protocol that you inject on top of genesis. Your block chain would have a tezos genesis block, then a block that activate your own version of bitcoin, and then the blocks would be similar to what you would find on the bitcoin block-chain.

Of particular interest to you, the protocol you inject can have entirely different on-chain notions (e.g., a TCG/CCG with no coins at all but a notion of ownership over cards) and different soft-updating mechanism (e.g., the new protocol can accept genesis-style updates (a “dictatorship” where a single person controls the protocol) or even no soft-updating mechanism at all (a “stale” protocol where you need to hard-fork if you want to make significant changes)).

For this use case (of starting your own chain with a different protocol), you might be better off cloning the git repository, doing some minimal clean up, etc. This is because the tezos binaries include the sources for all protocols that have been used on the chain (so you don't need to get them over the network even if you can).

You might be interested in the following blog post about how to write your own protocol: https://blog.nomadic-labs.com/how-to-write-a-tezos-protocol.html

The Everest team is pleased to announce the release of official OCaml bindings for all of our verified cryptographic algorithms, now available through OPAM as packages hacl-star and hacl-star-raw.

We provide bindings for the following:

• HACL*, a library of pure C algorithms
• Vale, a collection of optimized core assembly routines for maximum performance
• EverCrypt, an agile, multiplexing API with CPU auto-detection that brings together HACL* and Vale.

Our code is compiled from the F* programming language to C via the KReMLin compiler ("K&R meets ML"). We offer two OPAM packages:

• hacl-star-raw consists of low-level ocaml-ctypes bindings generated by KReMLin
• hacl-star is a hand-written OCaml idiomatic API that uses much more pleasant signatures, types and abstractions and is also safer, as it checks all static preconditions at run-time

We support AES{128,256}-GCM, Chacha20-Poly1305, Curve25519 / Ed25519, P256, MD5, SHA-{1,2,3} (all variants), Blake2 (s&b), HMAC/HKDF, and the HPKE and SecretBox high-level APIs. Some algorithms are optimized for Intel chips, notably AES-GCM – see https://hacl-star.github.io/Supported.html for full details.

General documentation about the project is available at https://hacl-star.github.io/index.html – sample code for the OCaml API is provided as part of the test suite https://github.com/project-everest/hacl-star/tree/master/bindings/ocaml/tests

This work was performed by Victor Dumitrescu from Nomadic Labs, one of the teams responsible for the core development of the Tezos blockchain.

Ahoy developers, few days ago I published a new ocaml library called nmea, which is essentially a parser for NMEA0183 sentences, a format for encoding instruments data in boats. There are many sentences, regarding GPS, compass data, wind, air pressure, water temperature, waypoints handling, ais, autopilot and more; at the moment the library is able to decode GPS sentences and compass data, but I'll implement more sentences in the spare time. I tested it with my boat GPS and with a gps usb dongle.

After that, I started a new tiny experiment called sail-gadgets, which is a Gtk program that elaborates and displays NMEA data received from various boat instruments (wind vane, autopilot, gps, radar, ais, etc). Sail-gadgets can be extended with "gadgets" modules, each one providing new functionalities and new tabs to the main interface.

Data from sensors are handled using React signals, so in every gadget we can compose data from various sensor to obtain new reactive values.

The gadgets I'm planning to write:

• dashboard: shows current position, speed, heading, tripdist, compass
• satview: shows current connected gps satellites (partially done)
• wind: shows wind indicator with true / apparent speed and direction
• radar: shows AIS and Radar targets in range
• mob: allows to drop a marker in the current position, and drive you to that point
• startline: helper for regatta start
• track: shows current track in a vector map

The hard thing in my opinion is writing new custom widget with cairo (compass, radar, and things like that).

Finally, the project is intended to run over gtk-broadway, so every html5 enabled device can access the application.

Hope there are some sailor here that want to join writing some gadgets :) Repos are:

• https://github.com/dakk/nmea
• https://github.com/dakk/sail-gadgets

I searched to find math library which is written in OCaml, but there are only few repositories.

I'd like to use some function like coefficient correlation, covariance, etc.

I found Lacaml but it seems not to support them.

Could you give some link if you know?

Hi, you can take a look at Owl : https://ocaml.xyz/

There are stat functions and also a lot more

There is also this one:
https://github.com/superbobry/pareto
GSL powered OCaml statistics library http://superbobry.github.io/pareto/0.2

And probably even some more:

opam search statistic
# Packages matching: match(*statistic*)
# Name            # Installed # Synopsis
[...]
gsl               --          GSL - Bindings to the GNU Scientific Library
oml               --          Math Library
owl               --          OCaml Scientific and Engineering Computing
owl-plplot        --          OCaml Scientific and Engineering Computing
pareto            --          GSL powered OCaml statistics library.
statsd-client     --          StatsD client library
[...]

I wonder, why there are no new OCaml books since 2014 year? Many books are published on Haskell, Scala, F# themes, but no OCaml. I think we need new books for learning and for rising interest in our beautiful language.

• There is an ongoing work for 2nd edition of Real World OCaml: http://dev.realworldocaml.org/.
• OCaml Scientific Computing is also ongoing: https://ocaml.xyz/book/.
• A Japanese book "コンピュータを操る", published in Feb. 2020 for beginners of programming, uses OCaml Blockly: https://www.saiensu.co.jp/search/?isbn=978-4-7819-1470-1&y=2020#detail.

A Chinese book OCaml语言编程基础教程 (an introduction to OCaml language programming) is published in 2018.

Here are links from many OCaml blogs aggregated at OCaml Planet.

• Every proof assistant: MMT