Another short post on pybind11 and how you can leverage its call guard directives to release the GIL and achieve true multithreaded parallellism in python.

Recently, I’ve been writing some C-extensions to speed up some performance critical code in Python and while trying to setup pybind11 on a project managed with Poetry I realized there was no clear documentation on how to setup the build process correctly. Here’s a short write-up as in how to sort it out.