Omyyyy
Nuance README claims
The readme states:
generating native code that can run as a standalone executable from Python code has never really been done
But the nuitka (https://nuitka.net/) project has been doing exactly that, for C, for a long time.
Nuitka is stable, support all Python stdlib, syntax and most 3rd party projects, including compiled extensions like numpy. It transpiles to C then compile to several targets, one of them being a native standalone executable.
Speed gain can go up to X4 python perf.
Pycom of course, gets 20-30x, but because it doesn't support the most dynamic python features which are what makes python slow: site imports, zip imports, heterogeneous types, monkey patching for built-ins, dundder methods, etc. As pycom progresses, you'll note the gain to go down significantly.
This is an awesome project, but it would gain from having a more nuanced README.
Tried explaining some of these to the author back when he released it but I guess some things are just better experienced than heard/read.
Unfortunately, right now authors cannot properly manage open-source project
Incompetence is blinding, I guess ;0
This will definitely not help and isn't productive Although I know one thing, we cannot decide or act for the author
I'm not sure being rude to the author is going to help.
Telling hard facts can sometimes seem like being rude.
Personally, I have no problem with the author or project, in fact I see quite some potential here but at the same time, it's important to be realistic about the things @ksamuel has mentioned.
There are a lot of python compilers. It doesn't mean someone should not try to make another one. But claims about the possibilities of this only a nascent project are misleading.
@ksamuel: FWIW I couldn't get Nuitka to compile a simple print("Hello world") to less than 200KB, which executes about 20x slower than the 65KB executable produced by Pycom. I don't know if Pycom can maintain that speed advantage as it implements more python features, but right now the ability to produce a small fast executable seems to be a big advantage of Pycom over Nuitka.
@phodge Cython (industry standard as it's 20+yo and used in a lot of popular packages) produces executables that are faster from 2 to 1000 times vs just running Cpython a.k.a. Python It supports everything you need in python and beyond This should be interested to you
@phodge py2many takes the same approach as pycom and has been around for 7+ years under different names. Giving up python C API has a number of benefits like you observe.
