Pyxell [pixel] aims to combine the best features of different programming languages, pack them into a clean and consistent syntax, and provide the execution speed of native machine code.
It draws mainly from Python, C++, C#, and Haskell, trying to avoid common design flaws that have been nicely described in this blog post.
To do:
Python 3.6+
C++17 compiler: GCC 7+ or Clang 5+
python pyxell.py program.px
If the program is valid, program.cpp
file and program.exe
executable will be created in the same folder,
and it will be automatically executed (unless you add the -n
flag).
Otherwise, errors will be displayed, pointing to the erroneous code location.
By default, gcc
command is used to compile the code.
You can pick a different compiler using the -c
parameter.
The executable is not optimized by default.
You can set an optimization level with the -O
parameter, e.g. -O2
.
This will make the program run faster, but also make the compilation slower.
In order to speed up the compilation, you can precompile the C++ header (lib/base.hpp
) by first running the script with the -p
flag.
Note that the precompiled header is compatible only with the same C++ compiler, optimization level, and Pyxell version.
Use -s
to skip the compilation step and obtain transpiled C++ code with all headers included,
ready for manual compilation (with -std=c++17
).
To see all command line options, use -h
.
pip install -r test/requirements.txt
python test.py
Tests are divided into good (supposed to compile and run properly) and bad (should throw compilation errors).
By default, the whole C++ code for valid tests is merged, so that only one file is compiled, which is faster than compiling hundreds of files individually, even using multiple threads. Total execution time (with default settings) should be around 30-60 seconds.
If, however, the script fails with an error like this: too many sections
/ file too big
(seen with GCC 7.2 on Windows), or there is another compilation error that is hard to decipher,
then you might need to add the -s
flag so that each test is compiled separately.
You can pass a path pattern to run only selected tests (e.g. python test.py arrays
).
To build the documentation from source, go to the docs
folder, run npm install
, then make
.
To start a documentation server locally, run pip install -r requirements.txt
and python server.py
in the same folder.
You can build a standalone compiler application using PyInstaller
.
Install PyInstaller
with pip
, then run make exe
.
An executable (not requiring Python to run) will be created in the dist/pyxell
folder.
There are only a few languages with indentation-based syntax. Some more or less similar to Pyxell are, in alphabetical order: