Skip to content

Latest commit

 

History

History
55 lines (39 loc) · 2.56 KB

README.md

File metadata and controls

55 lines (39 loc) · 2.56 KB
Raw

NPython

(Subset of) Python programming language implemented in Nim, from the compiler to the VM.

Online interactive demo by compiling Nim to Javascript.

Purpose

Fun and practice. Learn both Python and Nim.

Status

Capable of:

  • flow control with if else, while and for
  • basic function (closure) defination and call. Decorators.
  • builtin print, dir, len, range, tuple, list, dict, exceptions and bunch of other simple helper functions
  • list comprehension (no set or dict yet).
  • basic import such as import foo. No alias, no from, etc
  • raise exceptions, basic try ... except XXXError ... , with detailed traceback message. Assert statement.
  • primitive class defination. No inheritance, no metatype, etc
  • interactive mode and file mode

Check out ./tests to see more examples.

How to use

git clone https://github.com/liwt31/NPython.git
cd NPython
nimble c ./Python/python
./Python/python

Todo

  • more features on user defined class
  • builtin compat dict
  • yield stmt
  • better bigint lib

Performance

Nim is claimed to be as fast as C, and indeed it is. According to some primitive micro benchmarks (spin.py and f_spin.py in tests/benchmark/), although NPython is currently 5x-10x slower than CPython 3.7, it is at least in some cases faster than CPython < 2.4. This is already a huge achievement considering the numerous optimizations out there in the CPython codebase and NPython is focused on quick prototyping and lefts many rooms for optimization. For comparison, RustPython0.0.1 is 100x slower than CPython3.7 and uses 10x more memory.

Currently, the performance bottlenecks are object allocation, seq accessing (compared with CPython direct memory accessing). The object allocation and seq accessing issue are basically impossible to solve unless we do GC on our own just like CPython.

Drawbacks

NPython aims for both C and JavaScript targets, so it's hard (if not impossible) to perform low-level address based optimization. NPython currently relies on Nim GC. Frankly speaking it's not satisfactory.

  • The GC uses thread-local heap, makes threading nearly impossible (for Python).
  • The GC can hardly be shared between different dynamic libs, which means NPython can not import extensions written in Nim.

If memory is managed manually, hopefully these drawbacks can be overcomed. Of course that's a huge sacrifice.

License

Not sure. I think it should follow CPython license, but other Python implementations like RustPython use licenses like MIT.