Git Product home page Git Product logo

cowgol's Introduction

Cowgol

What?

Cowgol is an experimental, Ada-inspired language for very small systems (6502, Z80, etc). It's different because it's intended to be self-hosted on these devices: the end goal is to be able to rebuild the entire compiler on an 8-bit micro.

Right now it's in a state where you can build the cross-compiler on a PC, then use it to compile the compiler for a 6502 (or Z80) device, and then use that to (slowly) compile and run real programs on a 6502 (or Z80). It's theoretically capable of compiling itself but need memory tuning first. (And, realistically, bugfixing.)

The compiler itself will run on these architectures (as well as cross-compiling from a modern PC in a fraction of the time):

  • 6502, on a BBC Micro with Tube second processor; this is the only platform I've found which is big enough (as it gives me a real operating system with file streams and 61kB of usable RAM). (The distribution contains a simple emulator.)

  • Z80, on CP/M. (The distribution contains a simple emulator.)

  • Z80, on Fuzix; see http://www.fuzix.org. You'll need your own emulator, or real hardware to run on.

It will also cross compile for all of the above plus:

  • 6502, on the Commodore 64 (for ultra hackers only; email me).

  • Z80, on the ZX Spectrum (for ultra hackers only; email me).

Why?

I've always been interested in compilers, and have had various other compiler projects: the Amsterdam Compiler Kit and Cowbel, to name two. (The languages section of my website contains a fair number of entries. The oldest compiler which still exists dates from about 1998.)

Cowgol is a spinoff of the Amsterdam Compiler Kit --- thematically, although it shares no code. By dividing the task into many small, discrete units, it gets to do (if slowly) a job which machines this size shouldn't really be able to do. In many ways it's an exercise in minimalism, just like Cowbel, although in a different direction.

Where?

How?

We have documentation! Admittedly, not much of it.

  • Everything you want to know about Cowgol, the language; tl;dr: very strongly typed; Ada-like syntax; multiple return parameters; no recursion; nested functions.

  • An overview of Cowgol, the toolchain; tl;dr: eight-stage compiler pipeline; separate front-end and back-end; maximum RAM use: about 60kB; call graph analysis for efficient variable packing; suitable for other languages; written in pure Cowgol.

  • About the BBC Micro bootable floppy; tl;dr: crude, slow, not suitable for production use; requires a BBC Micro with 6502 Tube second processor although I recommend a BBC Master Turbo (mainly for the built-in editor); requires extreme patience as it takes eight minutes to compile a small program.

  • About the CP/M distribution; tl;dr: crude, slow, not suitable for etc; requires a Z80-based CP/M 2.2 or later system with at least 50kB of TPA.

  • About the Fuzix distribution; tl;dr: crude, slow, etc, etc. Requires a Fuzix system with a Normal™ Z80 ABI (i.e. not the ZX Spectrum) with at least 48kB of userspace.

processor although I recommend a BBC Master Turbo (mainly for the built-in editor); requires extreme patience as it takes eight minutes to compile a small program.

You will need some dependencies:

  • the Ninja build tool

  • Lua 5.2 (needed for the build)

  • the Pasmo Z80 assembler (needed to build part of the CP/M emulator)

  • the libz80ex Z80 emulation library (needed for the CP/M emulator)

If you're on a Debianish platform, you should be able to install them with:

apt install ninja-build lua5.2 pasmo libz80ex-dev

Once done you can build the compiler itself with:

ninja

You'll be left with a lot of stuff in the bin directory. The BBC cross compiler is in bin/bbc_on_native; the BBC native compiler is in bin/bbc. The BBC demo disk is in bin/bbcdist.adf. Likewise, the CP/M cross compiler is in bin/cpmz_on_native and the native compiler is in bin/cpmz.

To run the cross compiler, do:

./scripts/cowgol -a bbc_on_native -o cow.out \
    src/arch/bbc/lib/runtime.cow \
    src/arch/6502/lib/runtime.cow \
    src/arch/common/lib/runtime.cow \
    srctest.cow

You'll be left with a BBC Micro executable in cow.out. For the Commodore 64, substitute c64_on_native and src/arch/c64/... in the obvious places. For CP/M, substitute cpmz_on_native, src/arch/cpmz/..., and src/arch/z80/... in the obvious places. For Fuzix, substitute fuzixz80_on_native etc etc obvious places.

The first three input files should be always be the runtime library.

The compiler works by having a shared state, things.dat, which is read into memory by each stage, modified, and written out again on exit. Then there is the opcode stream, iops.dat, which is streamed through memory. Provided you have enough RAM for the things table you should be able to compile programs of unlimited size; you need 35kB for the things table to compile the compiler. This will fit, just, so it's theoretically possible to build the compiler on a BBC Tube, but it needs some other memory rearrangement before it's worth trying. (And, realistically, making the code smaller and more efficient.)

Special emulation bonus! Are on a Unix platform? Do you have *b-em or BeebEm? If so, there's a farm of symlinks on tools/vdfs which point at all the appropriate binaries and source files in the main ditribution, with .inf files already set of for you. You can point your VDFS root here and you should have a live setup just like the demo floppy, except much faster and with your changes saved. And without the risk of running out of disk space! Just remember to set your machine type to a BBC Master Turbo, and then crank the emulation speed for both the main computer and the Tube processor as high as they will go.

Even specialler emulation bonus! There's a built in emulator for CP/M *which will let you run Cowgol for CP/M out of the box using the farm of *symlinks in tools/cpm! After building Cowgol, do this:

$ bin/cpm -p a=tools/cpm/a -p b=tools/cpm/b/
a> submit compile

...and watch the fun! (If you get this running on real hardware, please let me know. I want to know how long it takes.)

Why not?

So you've tried one of the demo disks!

...and you've discovered that the compiler takes seven minutes to compile "Hello, world!". Does that answer your question?

There are a bunch of things that can be done to improve performance, but they all need memory. This isn't free, so I'll need to make things smaller, improve code sizes, make the generated code more efficient, etc.

But let's be honest; you're trying to compile a modern-ish language on a 2-4MHz device with 64kB of RAM. It's not going to be fast.

Who?

Cowgol was written, entirely so far, by me, David Given. Feel free to send me email at [email protected]. You may also like to visit my website; there may or may not be something interesting there.

License?

Cowgol is open source software available under the 2-clause BSD license. Simplified summary: do what you like with it, just don't claim you wrote it.

src/bbctube contains a hacked copy of the lib6502 library, which is © 2005 Ian Plumarta. See emu/bbctube/COPYING.lib6502 for the full text.

tools/cpm/a contains some tools from the original CP/M 2.2 binary distribution for the Imsai 1800, redistributable under a special license. See tools/cpm/a/!readme.txt for the full text.

cowgol's People

Contributors

davidgiven avatar dtrg avatar

Watchers

 avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.