With numpy 2.0 we run into several issues. One is solved here, but there are more changes regarding string/unicode handling that kill the continuous integration at the moment. Most of them I could not reproduce locally. As an intermediate solution I limited numpy to versions less than 2.0.
This pr fixes:
miniconda action by specifying the version to "latest"
coverage: 79.774% (-0.008%) from 79.782%
when pulling 34c4f49170feadcb0b59b343270ec3b14be80052 on jgrewe:numpy
into a89d84461a0b372b4155d3f67ef159a6a1147f1c on G-Node:master.
With numpy 2.0 we run into several issues. One is solved here, but there are more changes regarding string/unicode handling that kill the continuous integration at the moment. Most of them I could not reproduce locally. As an intermediate solution I limited numpy to versions less than 2.0.
This pr fixes:
pathlib
instead ofos
in test_docs@achilleas-k if you could spare a quick look ;)