Tutorial: Racket FFI, Part 2

:: Racket, FFI, tutorial, by Asumu Takikawa

This is part 2 of my tutorial on using the Racket FFI. If you haven’t read part 1 yet, you can find it here. Update: part 3 is also now available here.

Part 2 will continue with more Cairo examples. In this installment, I plan to go over some more advanced FFI hacking such as handling computed argument values, custom return arguments, and using C structs.

Tutorial: Using Racket’s FFI

:: Racket, FFI, tutorial, by Asumu Takikawa

Update: this post is now part of a series. Part 2 is here and part 3 is here.

I’ve seen several people ask for a tutorial on Racket’s foreign function interface (FFI), which allows you to dynamically load C libraries for use in Racket code. While I think the documentation for the FFI is quite good, it is a lot of information to process and the overview examples may be tricky to run for a beginner.

With that in mind, this blog post will provide a step-by-step tutorial for Racket’s FFI that requires minimal setup. All that you will need to follow along is a copy of Racket and ideally a DrRacket window.

Measuring GC latencies in Haskell, OCaml, Racket

:: garbage collection, latency, instrumentation, haskell, ghc, ocaml, racket, by Gabriel Scherer

James Fisher has a blog post on a case where GHC’s runtime system imposed unpleasant latencies on their Haskell program:

Low latency, large working set, and GHC’s garbage collector: pick two of three

The blog post proposes a very simple, synthetic benchmark that exhibits the issue — basically, latencies incurred by copy time — with latencies of 50ms that are considered excessive. I thought it would be amusing to reproduce the synthetic benchmark in OCaml and Racket, to see how other GCs handle this.

Without further ado, the main take-away are as follows: the OCaml GC has no issue with large objects in its old generation, as it uses a mark&sweep instead of copying collection, and exhibits less than 3ms worst-case pauses on this benchmark.

The Racket GC also does not copy the old generation, but its incremental GC is still in infancy (compared to the throughput-oriented settings which works well) so the results are less good. It currently suffer from a “ramp-up” effect that I will describe, that causes large pauses at the beginning of the benchmark (up to 120ms latency), but in its steady state the longest pause are around 22ms.

Please keep in mind that the original benchmark is designed to exercise a very specific workflow that exercises worst-case behavior for GHC’s garbage collector. This does not mean that GHC’s latencies are bad in general, or that the other tested languages have smaller latencies in general.

The implementations I use, with a Makefile encapsulating the logic for running and analyzing them, are available in a Gitlab repository:

Gradual Typing Across the Spectrum

:: gradual typing, PI meeting, by Asumu Takikawa

Instead of being Pythonistas, Rubyists, or Racketeers we have to be scientists. — Matthias Felleisen

Yesterday we hosted a PI meeting for the Gradual Typing Across the Spectrum NSF grant, gathering researchers from a number of institutions who work on gradual typing (the meeting program can be found here). In case you aren’t familiar with gradual typing, the idea is to augment dynamically typed languages (think Python or Ruby) with static type annotations (as documentation, for debugging, or for tool support) that are guaranteed to be sound.

Gradual typing is these days a fairly popular area, but the research can seem fragmentary because of the need to support idiosyncratic language features. One of the points of the meeting was to encourage the cross-pollination of the key scientific ideas of gradual typing—the ideas that cross language and platform barriers.