34a83839f5
be valid utf-8 strings |
||
---|---|---|
client | ||
contrib/vim | ||
core | ||
dsttest | ||
gsttests | ||
include/dst | ||
junkyard | ||
libs | ||
unittests | ||
.gitignore | ||
.travis.yml | ||
2 | ||
LICENSE | ||
Makefile | ||
README.md |
dst
dst is a language and vm that is small and embeddable, has metaprogramming facilities, can interoperate with C, and has enough features to make it a useful general purpose programming language. It is a variant of Lisp with several native useful datatypes. Some of the more interesting and useful features are first class functions and closures, immutable and mutable hashtables, arrays, and bytebuffers, macros (NYI), tail-call optimization, and continuations (coroutines, error handling). The runtime and bootstrapping compiler are written in C99, but should eventually be completely compatible with C89 compilers.
As of July 2017, still WIP. While the basic runtime is in place, as are many native functions, several important features are still being implemented and defined, like the module system and macros.
There is a repl for trying out the language, as well as the ability to run script files. This client program is separate from the core runtime, so dst could be embedded into other programs.
Compiling and Running
Clone the repository and run:
make run
To build the runtime and launch a repl.
Basic programs
To run some basic programs, run the client with one argument, the name of the
file to run. For example, the client that is built will be located in the
client directory, so running a program script.dst
from the project directory
would be
client/dst script.dst
You can also use the --help
option to see more usage information for the vm.
Running Tests
Simply run make test
to run the currently minimal test suite.
Todo
- Documentation, with the caveat that things may change.
- Use serialization to allow creation of bytecode files that can be loaded. This includes defining a file format for the bytecode files. This mostly done. The byte code serialization could also be useful for the module system for loading artifacts.
- Pattern matching/regex library, as well as string formatting functions. This would also be useful for implementing better error messages.
- Better error messages, expecially for compilation. This probably means string formating functions.
- Macro/specials system that happens before compilation
- Module system. Something similar to node's require.
- Change name (dst is the name of many projects, including GNU Smalltalk). Maybe make logo :)?
- Change C API to be stack based for fewer potential memory management problems. This could mean making current C API internal and use separate API externally.
- Store source information in parallel data structure after parsing
- Use source information during compilation
- Use Lua style memory alocator backend C API (one single function for allocating/reallocating/freeing memory).
- More builtin libraires.
- Fuzzing