-
Notifications
You must be signed in to change notification settings - Fork 299
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
How do I install cq in a conda env using pip #1236
Comments
Weird use case. If you can build a wheel, you can also build a conda package and install the deps from the env file. BTW: there is a PR in progress for arm64 OCP for the ocp-feedstock If you really need pip, maybe this will work:
|
What do you mean with "weird use case". I and some others do have arm64 Macs and want to work with CadQuery. I don't see the weird part of this use case. The wheel I built might have some issues (and I also had to create wheels for nlopt and casadi for arm64). While they do work with some simple tests, I don't want to publish them - and I have no clue about conda packaging. The approach up to the last commit was a good workaround until the OCP package exists. And agreed, as soon as there is an OCP arm64 conda package it is a non issue any more. Nevertheless, until then a workaround is needed. Maybe this is anyhow a good time to leave the conda ecosystem and move over to poetry .... I'll give your suggestion a try tomorrow, thanks. |
@bernhard-42 You can probably create your own local copy of the |
"Weird" was about mixing package managers and not the platform itself. BTW both casadi and nlopt are avaiable on conda-forge for osx-arm64. |
Got it, thanks @jmwright and @adam-urbanczyk Not happy with mixing myself. However, I only had a private arm64 wheel for cadquery_ocp (otherwise I couldn't work with cadquery on my machine). And yes I could maintain my own environment.yml ... Btw. what is the reason for preventing dependencies being installed via pip in a conda env. Usually, this will be controlled on the client side by ˋpip install --no-depsˋ and not prevented in the package, if I understand the python packaging approach correctly. |
env -u CONDA_PREFIX_1 pip install git+https://github.com/cadquery/cadquery.git works. But this is really looks a bit "weird" now ;-) |
See #1228 and pypa/pip#9948 |
makes sense, thanks |
@bernhard-42 would you be able to test this experimental OCP build https://anaconda.org/cctbx-dev/ocp with CQ on your arm64 mac? See conda-forge/ocp-feedstock#23 for context. |
Sure, will test it tomorrow and let you know |
@adam-urbanczyk I tried to make some kind of end-to-end test with vtk 9.2.2 - seems to work: |
For reference, the environment setup: mamba create -n ocp-test python=3.8
mamba activate ocp-test
mamba install -c cctbx-dev ocp vtk pyvista panel jupyterlab |
Thanks @bernhard-42 ! You can use osx-arm64 ocp from conda-forge now. |
Awesome! Thank you |
The latest commit prevents from dependencies being installed when in conda env.
On a M1 mac I have my own OCP wheel sitting on a private pypi server, called "banach"
My install process was for an existing conda env:
mamba create -n cq-3.9 python=3.9 ipython jupyterlab matplotlib numpy pandas # for example, plus any other libs pip install --pre --trusted-host banach --extra-index-url http://banach:8080/ git+https://github.com/CadQuery/cadquery.git
This now doesn't install any dependencies any more.
Why do I start with conda? Because mambaforge provides M1 arm64 packages that still don't exist sometimes on pypi.
If I fall back to the commit before the change in
setup.py
it works againThe text was updated successfully, but these errors were encountered: