Conversation
…e with function to call rust implementation
…Rust idomatic way
|
Hi Kristof (@schroedk), looks like you're making amazing progress (and I am also told that performance looks pretty promising? 😄 ). Let me know once I should start to review + in case you want me to give you the rights to run the CI yourself? |
…ic way, add simple test
|
Hi Alexander (@s3alfisc ), I modified the code such that the rust code will be packaged into the wheel for upload to pypi (Have a look at pyfixest.core). You can use What is still missing:
Next week I will be on vacation, but I will text you, when I am back. I will give you access to my fork, but you can also merge into your feature branch first if this is more convenient. |
|
Awesome, thank you! I'll review over the next days and then likely merge into my branch (unless you'd prefer otherwise?). Plus I'll tackle the docs. I'll also take a look at package building and will play around a bit with the new pixi build feature. Have a great vacation and thanks so much! |
|
Huh just took a look at the benchmarks, these are pretty spectacular. 10% reduction for the demeaning algo is super good! Cool! 😎 |
This PR implements the packaging of the rust components as the subpackage
pyfixest.core._core_impl.It adds dependency to pytest-benchmark for comparing the different backend implementations while testing.
So far, rust implementations are on par or faster than numba compiled code.





