Closed rdebokx closed 3 years ago
That's a good point. I had quite some issues with using Python 3.7 in combination with boost, as also documented in #2 and https://github.com/boostorg/boost/issues/462. The members of the boost mailing list advised me to use the Python version that was shipped with the Ubuntu version I was using. for Ubuntu 18.04 that was 3.6 and for Ubuntu 20.10 that's 3.8.
In make LIB_BOOST_PYTHON=-lpython3.6m LIB_BOOST_NUMPY=-lboost_numpy3
I indeed overwrite the variables as they are declared in Makefile-variables.mk
. This way you don't need to manually change anything in Makefile-variables.mk
in case you want to run GP-GOMEA using Docker.
Your request got me thinking though; with the understanding I gathered by now about boost and linking it to python and numpy, I was able to use Ubuntu 20.10 instead and use Python 3.8; this is indeed more future-proof. I still think it's good to declare these versions in the make
command in the Dockerfile; this way a change in Makefile-variables.mk
won't break the Docker building process.
Please let me know what you think!
@rdebokx perfect, it makes a lot of sense. Thanks for updating it to Ubuntu 20.04 with python3.8. I really appreciate this, thank you!
This PR implements support for running GP-GOMEA in a Dockerized setup. The contents of the
Dockerfile
are based onm_ubuntu
, with some minor tweaks to avoid issues with linking Python and Numpy.This enables users to run GP-GOMEA on any operating system that Docker can run on. For some operating systems this may have an (yet unspecified) performance impact.
Fixes #2