The Mu shell, compiling down to a subset of 32-bit x86 machine code, then to a Linux ELF binary, packaged up with just a Linux kernel and nothing else, running on a Linux console emulated on Qemu, on a Thinkpad T420s running 64-bit Linux.

Just another 27 million lines of C to take out (Linux kernel), and I'll have a decent computing stack.

---

I'm also thinking today about what it would take to fork Mu for the Raspberry Pi. It uses ARM processors, a 64-bit instruction since the Pi 3 which was released in 2015. The 64-bit ISA looks quite nice, but it's incompatible with the Pi 1 and Pi 2. Contrast x86, where 64-bit is quite compatible with 32-bit, but grotesque as a result, with bits for a register scattered across multiple bytes. So I might make the opposite choice to x86 and support just the 64-bit ISA. Thoughts?

https://developer.arm.com/documentation/den0024/a

permalink

Screenception

Visualizing programs with side-effects in a postfix shell with a live-updating text-mode environment. Built all the way up from machine code without any dependencies (except an x86 processor and Linux kernel).

https://archive.org/details/akkartik-2min-2020-11-12

Project page: https://github.com/akkartik/mu

More context

permalink

In the last 3 days I've written 68 tests for error messages in the Mu compiler, adding up to about 4.5k lines of machine code. Utterly tedious, mind-numbing work with lots of copy-pasta. Just what the doctor ordered.

(https://github.com/akkartik/mu)

permalink

For a couple of years I've been building up a computing stack without metacircularity, where complex languages are consistently implemented in simpler languages.

For several months now I've been wrestling with a thorny problem in one corner of the core compiler which converts a safe language into unsafe machine code. Today I finally decided to stop agonizing over it, and write up the idea maze to the extent I've explored it.

https://github.com/akkartik/mu/issues/45

Comments and suggestions most appreciated! This is a fairly simple compiler as these things go, and I'd be happy to engage with anyone who wants to learn about these beasts in a realistic setting.

permalink

My little prototype is starting to look like a shell

https://archive.org/details/akkartik-2min-2020-10-27

Promising in some ways, but I'm not sure how to support concurrency. Currently each operation completes before the next. I could allow "pipe stages" to continue to share data after they drop file handles on the stack, but there are problems: how often we refresh, how we kill processes from past refreshes, how we visualize file handle contents.

(More details: https://github.com/akkartik/mu)

permalink

Beginnings of an experimental fork of Mu for 64-bit x86

Built in collaboration with @tekknolagi.

https://git.sr.ht/~akkartik/mu-x86_64

I'm not sure this is going to work out. Mu's syntax for 32-bit x86 machine code doesn't map cleanly to x86_64, for reasons outlined in the Readme. But the emulator works, and it has a nice regular subset of x86_64 including floating-point instructions. Comments and suggestions most appreciated.

Main project page: https://github.com/akkartik/mu

permalink

This week in my postfix language and live-updating environment, I worked on a way to start from a raw computation, and extract functions from it as naturally as possible.

https://archive.org/details/akkartik-2min-20201020

(More details: https://github.com/akkartik/mu)

permalink

My postfix language and its live-updating environment are starting to look promising. The environment can now expand multiple levels of function calls, laying out the state of the stack at each point. You can't edit a function at its call-site, but you can visualize its working in context.

https://archive.org/details/akkartik-2min-2020-10-10

(More details: https://github.com/akkartik/mu)

permalink

Detective story of the day

I've been stress-testing Mu's floating-point instructions for a few days using this ray-tracing tutorial.

It's been great; I've found 2 bugs so far.

Today I thought I found a third, in the floating-point reciprocal instruction.

Except it wasn't really. Read on.

permalink

I just finished adding floating-point to Mu. Took 4 days. Floating-point is where the ugliness of x86 really becomes apparent.

permalink