I am using nuitka to compile my python 3.9 project, generate some executable files and some *.so modules.
As my project start using numpy and pandas, and some other libraries. The nuitka compilation time become too long to wait.
Is there any methods to optimize nuitka compilation time ?
For example, can I exclude numpy and pandas from packaging into executable files ?
Since it is likely, you are reusing exactly the same versions of pandas and NumPy code (along with dependencies), you should be able to generate the C code for these only once and reuse. On Linux this is typically done using tools like ccache. You may want to look into that.
I can see there is caching for the Visual Studio (Windows) world developed by Nuitka developers: clcache so you are likely to be a good installing ccache to be used by gcc (which transparently is picking up compiled objects - *.o files - when the code does not change).
