KC Sivaramakrishnan Research Associate @ OCaml Labs

Continuous Benchmarking & Call for Benchmarks

Over the past few weeks, at OCaml Labs, we’ve deployed continuous benchmarking infrastructure for Multicore OCaml. Live results are available at http://ocamllabs.io/multicore. Continuous benchmarking has already enabled us to make informed decisions about the impact of our changes, and should come in handy over the next few months where we polish off and tune the multicore runtime.

JFP Special Issue on Algebraic Effects and Handlers

Andrej Bauer and I are editing a special issue of JFP on the theory and practice of algebraic effects and handlers. The CfP is below.


A deep dive into Multicore OCaml garbage collector

I recently gave a talk on the internals of multicore OCaml GC at Jane Street offices in NYC. The slides from the talk are available online. But I felt that the slides alone aren’t particularly edifying. This post is basically the slides from the talk annotated with notes.

(Monadic) Reflections on Concurrency

We recently published a paper on concurrent system programming with effect handlers. In this paper, we show that with the help of effect handlers, we could express in direct-style, various interactions of a concurrent program with OS services that typically require callbacks. The question is what do we do about legacy code that uses monadic concurrency libraries such as Lwt and Async. Surely a wholesale rewrite of all Lwt and Async code is a no go. This post is an exploration of some ideas to make Lwt and Async compatible with direct-style code.

Building and Publishing an OCaml Package: Q1 2017

One of the key indicators of maturity of a language ecosystem is the ease of building, managing and publishing software packages in that language. OCaml platform has made steady progress in the last few years to this end. While OPAM simplified package (and compiler) management, the developing and publishing packages remained a constant pain point. This situation has remarkably improved recently with the Topkg and Carcass. This post provides a short overview of my workflow for building and publishing an OCaml package using Topkg and Carcass.

Ezirmin : An easy interface to the Irmin library

Ezirmin is an easy interface over the Irmin, a library database for building persistent mergeable data structures based on the principles of Git. In this post, I will primarily discuss the Ezirmin library, but also discuss some of the finer technical details of mergeable data types implemented over Irmin.

Behavioural types

Behavioural types such as session types, contracts and choreography describe the behaviour of a software entity as a sequence of operations on a resource such as a communication channel, web service session or a file descriptor. Behavioural types capture well-defined interactions, which are enforced statically with the help of type system machinery. In this post, I will describe a lightweight embedding of behavioural types in OCaml using polymorphic variants through a series of examples. The complete source code for the examples is available here.

Lock-free programming for the masses

Efficient concurrent programming libraries are essential for taking advantage of fine-grained parallelism on multicore hardware. In this post, I will introduce reagents, a composable, lock-free concurrency library for expressing fine-grained parallel programs on Multicore OCaml. Reagents offer a high-level DSL for experts to specify efficient concurrency libraries, but also allows the consumers of the libraries to extend them further without knowing the details of the underlying implementation.

Armed with Reason

This is a short tutorial on how to build Reason apps for an ARM target with the help of Docker. I am using Docker for Mac, which is still under beta program. Using Docker for development has two important advantages over traditional cross-compilation. First, the Reason toolchain comes packaged as a Docker image and hence no local installation is necessary. Secondly, cross-compilers are often tricky to get right. Docker for Mac comes with multiarch support and hence removes the need for traditional cross-compilation.

Profiling the stack

In the last post, I described a flat allocation profiler for OCaml 4.02 bytecode interpreter. In this post, I’ll describe further developments which add support for call stack information and better location information. Lets dive straight to the usage: