Loko Scheme: bare metal optimizing Scheme compiler

6 days ago (scheme.fail)

> Loko’s runtime uses concurrency based on Concurrent ML.

That one could be a big deal.

https://scheme.fail/manual/loko.html#Concurrency

The docs don't mention channels and say that fibres are built on limited continuations (call/cc) which suggests missing some of the clever stuff in CML (e.g. that threads deadlocked on a channel get garbage collected)

  • While being heavily inspired guile-fibers, it seems to not actually be parallel.

    Anyway, I would say it is actually the nicest way to write concurrent programs. It supports you and helps you to not shoot yourself in the foot, while also staying out of the way.

I also like that it runs on bare metal. I wonder is it builds and runs on macOS? (I am on a mobile device and can’t check it myself.)

Off topic, but I love the design of the linked web site.

  • After skimming the docs and the site, I suspect this is x86-only for now. No mention of ARM or other processor architectures and the listed hardware support suggests a PC hardware target. You could probably cross compile on an Apple silicon Mac and run under emulation (e.g., QEMU).