... a toy scheme interpreter
This is very incomplete, probably quite buggy and a toy anyway. That said:
- Runs basic scheme code, e.g. all the samples from "The Little Schemer" and a lot more
- Direct interpreter in plain C, but with tail calls
- Compiler stage implemented in scheme (very simple and ugly)
- Precise and generational garbage collector (slow anyway)
- Strings, vectors, booleans, 55-bit ints (!) and 32-bit floats
- Short string and symbol optimisations to make them non-heap objects
- Caching of symbol lookups
- Dynamic bindings
Over time I would like this to be a complete and correct R7RS implementation,
while still being built as a direct interpreter. The next planned steps are in
the ToDo List, and there are a lot of XXX fixmes in the code.
After checking out the repository, you need to pull the dependencies that reside in submodules:
git submodule update --init --recursive
After that you can build keiryaku and run the built-in tests:
make
make test
This is known to work on 64-bit linux with GCC 8.3 as well as CLang 7.0, if you have any interesting build experiences please let me know.
Running keiryaku directly will give you a basic REPL with line editing and history:
./keiryaku
But it can of course also read scheme code from STDIN:
./keiryaku < test.ss
You can skip the REPL part and execute a scheme file directly:
./keiryaku test.ss
Which also means that you can put a shebang line in a scheme script and execute it directly if your keiryaku is anywhere in your $PATH, e.g.:
#!/usr/bin/env keiryaku
(display "hello world!")
(newline)
There are a handful of optional cmdline arguments, but they are only really
really required for development, e.g. to switch on interpreter debugging. You
can get a list with --help.
Keiryaku is licensed under the Artistic License 2.0.
Please let me know what you think: Robert Lemmen [email protected]