Muon is a modern low-level programming language, inspired by C, C#, Go, Rust and Python.
To stay up-to-date on Muon, consider following me on Twitter.
-
Strongly, statically typed.
-
Data oriented. Just functions, structs and enums. NO: classes, inheritance, properties, etc.
-
No runtime. Lack of a runtime makes the language simpler, reduces application startup latency and makes it easy to use Muon code from other languages.
-
Extremely minimal core. A language should not dictate dependencies. There is a standard library, but it is completely optional.
-
High performance. Strive for parity with C.
-
Flexible memory management. Programmers can switch between allocators dynamically and can define their own allocators.
-
Avoid common memory safety pitfalls. Memory is initialized to zero. Array bounds are checked (can be turned off where needed).
-
No undefined behavior. Undefined behavior can lead to various, hard-to-spot, bugs. In Muon, all behavior, including platform-specific behavior, is defined.
-
Ergonomics matter. Programmers spend a lot of time working with a language, so ergonomics are important. Muon has:
- Type inference for function return values and locals
- Generics
- Order independent declarations
- Newline as statement separator
- Uniform function call syntax
- Reference type notation
- Namespaces
-
Fail fast. Usually, error reporting/handling happens via return values. For unrecoverable errors and errors that a caller is not prepared to handle, Muon provides abandonment.
-
Small(-ish) language. Strive for a small, simple language. Having fewer ways to do something encourages a more consistent, focused ecosystem.
-
Fast & snappy tools. Provide tools centered around fast feedback and improving program understanding. E.g.: language server, REPL, hot reloading, debuggers, profilers.
A glimpse of Muon:
Array {
countOccurrences(items Array<T>) {
map := Map.create<T, int>()
for items {
count := map.getOrDefault(it)
map.addOrUpdate(it, count + 1)
}
return map
}
}
main() {
::currentAllocator = Memory.newArenaAllocator(4096)
s := "How much wood could a wood chuck chuck if a wood chuck could chuck wood?"
freq := s.split(' ').countOccurrences() // Equivalent to: Array.countOccurrences(ref string.split(s, ' '))
for e in freq {
Stdout.writeLine(format("word: {}, count: {}", e.key, e.value))
}
}
To get started with Muon, see getting started. To learn more about the language, see Muon by example.
Also, check out the roadmap to see which features are coming up.
- Compiler: implements error recovery and has column accurate error reporting, which should make for a pleasant command line experience.
- Language server: provides interactive language features, such as symbol search, go to definition and as-you-type diagnostics.
- VSCode extension: provides syntax highlighting and language features via the language server.
- ffigen: takes a .c/.h file and generates a corresponding .mu file with foreign interface declarations (a.k.a. bindings).
- More tools are planned.
Compiler backend. The compiler currently outputs C code. This means that we inherit C's undefined behavior model, which goes against the goals listed above! An LLVM backend is in the works which will avoid any undefined behavior.
Performance. The compiler is pretty fast. A basic benchmark -- compiling the Muon compiler (which is itself written in Muon), which is ~12K lines of code, on a 4Ghz core i7 -- shows a compilation speed of ~0.5 million lines/second. The compiler is single threaded right now and there's lots of room for further improvement. One major caveat: after the Muon compiler has finished, a C compiler still needs to run to generate the final binary, which usually takes up the most time. The LLVM backend will (hopefully) reduce this.
Supported platforms. Muon aims to target all popular platforms, including Windows, macOS, Linux, iOS and Android. 64-bit architectures are the main focus (more details).
To stay up-to-date on Muon, consider following me on Twitter.