Hi,
when running the “tutorial informed search” notebook, I get an error “ModuleNotFoundError: No module named ‘commonroad_dc’”. I know that there are similar errors but they occurred with “commonroad_dc.pycrcc” so not exactly the same and the tips there did not help me.
Could you please tell me what your OS version is so that I could check if there is an easier way of installing a modern version of boost on your machine (using a package manager)?
We have no Mac computer with the M1 chip at the Chair but the library builds fine on macOS Catalina and Monterey with the intel chip.
Compiling boost from source could solve this particular problem. But there could be other problems on the M1 cpu: we are not entirely certain that all python packages are available for the M1 chip. Please let us know how it goes.
I was not able to solve it yet. I built boost from source but it did not solve the issue. But I am probably not the only one with M1. I will try to redo the whole installation guide to see if an error occured somewhere else.
because they are independent from all other steps. In addition, build.sh always cleans up temp files before the new build.
It is not immediately clear to me if the boost is installed into the correct folder or if it is not picked up due to some apple m1 build peculiarities. It depends on whether there is the file foreach.hpp in the folder /usr/local/include/boost.
This is confirmed. Drivability Checker can currently not be built on macOS on m1 CPU. The docker version should possibly work fine.
Magnus and I could try to make it compatible with apple silicon this week. Drivability Checker could probably be made to support apple silicon, but I’m not certain if all necessary python packages are available for apple silicon.
If you have an apple m1 cpu, you could try to use docker instead of installing commonroad-search directly. There is no guarantee the docker image would work on m1 silicon but it is possible that it will. Many intel-based docker images do work on m1 silicon.
The reason why it does not build on m1-based macOS is related to the fact that homebrew under this operating system has different include and library paths for m1 silicon and for intel silicon. The two directions we would try are as follows.
Use a binary wheel built on an intel-based mac machine through Rosetta2 installing it with Anaconda python 3.8.
Build the library natively for m1 silicon (arm64 architecture) after its modifications and check if all the required python dependencies are available in the special version of miniforge for m1 silicon (python 3.9).
Hi, I just met the same problem with command ./build.sh -w .i, using Apple silicon. If there is an update for installation? I have also tried docker and Ubuntu system (Arm64) based on the platform UTM, these don’t work.
Thanks:)