NeuroTechX / EEG-ExPy

EEG Experiments in Python
https://neurotechx.github.io/EEG-ExPy/
BSD 3-Clause "New" or "Revised" License
437 stars 124 forks source link

Conda 1st time install conflicts with existing python install #37

Closed retiutut closed 3 years ago

retiutut commented 3 years ago

I tried:

conda create -n "eeg-notebooks"

conda activate "eeg-notebooks"

conda install git

conda install pip

git clone https://github.com/NeuroTechX/eeg-notebooks

cd eeg-notebooks

pip install -r requirements.txt

but encountered errors on the last step, since it tried to use my existing python 2 install on my Mac.

@m9h helped me using a different set of instructions.

retiutut commented 3 years ago

I found some instructions that seem similar to what helped me in a currently unused markdown file:

https://github.com/NeuroTechX/eeg-notebooks/blob/master/doc/getting_started/setup_overview.md#2-clone-eeg-notebooks-and-install-dependencies

ErikBjare commented 3 years ago

This is very surprising to me. If you ran conda activate "eeg-notebooks" in the same shell as you later ran pip then shouldn't that override any system pip install by placing the conda env bin path early in the $PATH env var? (that's at least how plain venv and similar behave)

Curious how that wasn't the case. What was the difference that fixed it for you?

retiutut commented 3 years ago

@ErikBjare This only happened once, and I would like to try again with a fresh start on a different Mac that already has Python, but not Conda, installed.

I'll report back and close this issue if I am unable to replicate.

retiutut commented 3 years ago

Curious how that wasn't the case. What was the difference that fixed it for you?

conda create -n "eeg-notebooks" Python=3

Specifying the python version when creating conda environment was the crucial step that made it work with no issues.

ErikBjare commented 3 years ago

@retiutut Aaah, that makes a lot of sense. Surprised to see conda not pick Python 3 by default.

ErikBjare commented 3 years ago

@retiutut I made a docs PR #51 that should help avoid similar issues in the future.