Closed stargaser closed 1 year ago
Just updating my local miniconda didn't give me the "incompatible architecture" problem. I'm assuming that issue was on an Intel Mac. Was it with just the latest Python 3.10 version or was it also seen with a fresh install of the latest Python 3.9 version?
Anastasia encountered this on her Intel Mac with the latest version which must be Python 3.10. The second newest worked which must be the Python 3.9 version.
latest version which must be Python 3.10
not python 3.11?
I would still suggest trying adding the different solver, that shouldn't add confusion, but would provide a much faster experience.
I would still suggest trying adding the different solver, that shouldn't add confusion, but would provide a much faster experience.
I won't object to that, though I think it obligates some of us to do a little testing of the new instructions. Should we do that here, or should I merge this and look for a proposed change in another PR?
FWIW, I haven't been able to reproduce the miniconda error. Obviously there could be a lot of variables, but both of these downloaded (with correct SHA256 hash), installed and worked on my Intel Mac (MacOS 12.6.6 Monterey):
Miniconda3-py310_23.3.1-0-MacOSX-x86_64.pkg
Miniconda3-py310_23.3.1-0-MacOSX-x86_64.sh
This does make me wonder if the problem might not be very widespread.
I'll go ahead and merge this. Any new instructions can come in a new PR.
Okay, I've updated to remove mamba and add in the libmamba solver. See if this makes sense.
Oops, never mind, it's been merged, I missed that. Let's pu the change in another PR.
check_env.py
as it is not installed with Jupyterlab 4.0.Fixes #117