You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Docs: configure conda to use only conda-forge (#5569)
Requested by @ax3l:
> We might also need to rework our conda docs, to change the default
channel priorities to only use `conda-forge`. By default, many people
have the `anaconda` repos in them too and, although we do not use them,
this can break the build in this situation: some labs now block them in
their firewall due to recent events, and then the installer just aborts
instead of trying the next channel in the list. Very easy to do, I think
we just need to prepend `conda-forge` in the channels (instead of the
CLI) or add a guide to install Conda/Mamba from Conda-Forge/Minimamba
instead of Anaconda.
This PR adds instructions to:
1. add `conda-forge` to the top of the channel list;
2. set the channel priority to strict.
Hopefully this should be enough?
---------
Co-authored-by: Axel Huebl <axel.huebl@plasma.ninja>
Copy file name to clipboardexpand all lines: Docs/source/install/dependencies.rst
+12-13
Original file line number
Diff line number
Diff line change
@@ -52,27 +52,26 @@ For all other systems, we recommend to use a **package dependency manager**:
52
52
Pick *one* of the installation methods below to install all dependencies for WarpX development in a consistent manner.
53
53
54
54
55
-
Conda (Linux/macOS/Windows)
56
-
---------------------------
55
+
Conda-Forge (Linux/macOS/Windows)
56
+
---------------------------------
57
57
58
-
`Conda <https://conda.io>`__/`Mamba <https://mamba.readthedocs.io>`__ are cross-compatible, user-level package managers.
58
+
`Conda-Forge<https://conda-forge.org/download/>`__ is a repository for cross-compatible, user-level packages.
59
59
60
60
.. tip::
61
61
62
-
We recommend to configure your conda to use the faster ``libmamba`` `dependency solver <https://www.anaconda.com/blog/a-faster-conda-for-a-growing-community>`__.
62
+
We recommend to deactivate that conda self-activates its ``base`` environment.
63
+
This `avoids interference with the system and other package managers <https://collegeville.github.io/CW20/WorkshopResources/WhitePapers/huebl-working-with-multiple-pkg-mgrs.pdf>`__.
63
64
64
65
.. code-block:: bash
65
66
66
-
conda update -y -n base conda
67
-
conda install -y -n base conda-libmamba-solver
68
-
conda config --set solver libmamba
67
+
conda config --set auto_activate_base false
69
68
70
-
We recommend to deactivate that conda self-activates its ``base`` environment.
71
-
This `avoids interference with the system and other package managers <https://collegeville.github.io/CW20/WorkshopResources/WhitePapers/huebl-working-with-multiple-pkg-mgrs.pdf>`__.
69
+
In order to make sure that the conda configuration uses ``conda-forge`` as the only channel, which will help avoid issues with blocked ``defaults`` or ``anaconda`` repositories, please set the following configurations:
72
70
73
71
.. code-block:: bash
74
72
75
-
conda config --set auto_activate_base false
73
+
conda config --add channels conda-forge
74
+
conda config --set channel_priority strict
76
75
77
76
.. tab-set::
78
77
@@ -104,19 +103,19 @@ For OpenMP support, you will further need:
104
103
105
104
.. code-block:: bash
106
105
107
-
conda install -c conda-forge libgomp
106
+
mamba install -c conda-forge libgomp
108
107
109
108
.. tab-item:: macOS or Windows
110
109
111
110
.. code-block:: bash
112
111
113
-
conda install -c conda-forge llvm-openmp
112
+
mamba install -c conda-forge llvm-openmp
114
113
115
114
For Nvidia CUDA GPU support, you will need to have `a recent CUDA driver installed <https://developer.nvidia.com/cuda-downloads>`__ or you can lower the CUDA version of `the Nvidia cuda package <https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#conda-installation>`__ and `conda-forge to match your drivers <https://docs.cupy.dev/en/stable/install.html#install-cupy-from-conda-forge>`__ and then add these packages:
116
115
117
116
.. code-block:: bash
118
117
119
-
conda install -c nvidia -c conda-forge cuda cuda-nvtx-dev cupy
118
+
mamba install -c nvidia -c conda-forge cuda cuda-nvtx-dev cupy
120
119
121
120
More info for `CUDA-enabled ML packages <https://twitter.com/jeremyphoward/status/1697435241152127369>`__.
Copy file name to clipboardexpand all lines: Docs/source/install/users.rst
+12-13
Original file line number
Diff line number
Diff line change
@@ -38,36 +38,35 @@ If want to use WarpX on a specific high-performance computing (HPC) systems, jum
38
38
39
39
.. image:: conda.svg
40
40
41
-
Using the Conda Package
42
-
-----------------------
41
+
Using the Conda-Forge Package
42
+
-----------------------------
43
43
44
-
A package for WarpX is available via the `Conda <https://conda.io>`_ package manager.
44
+
A package for WarpX is available via `Conda-Forge<https://conda-forge.org/download/>`__.
45
45
46
46
.. tip::
47
47
48
-
We recommend to configure your conda to use the faster ``libmamba`` `dependency solver <https://www.anaconda.com/blog/a-faster-conda-for-a-growing-community>`__.
48
+
We recommend to deactivate that conda self-activates its ``base`` environment.
49
+
This `avoids interference with the system and other package managers <https://collegeville.github.io/CW20/WorkshopResources/WhitePapers/huebl-working-with-multiple-pkg-mgrs.pdf>`__.
49
50
50
51
.. code-block:: bash
51
52
52
-
conda update -y -n base conda
53
-
conda install -y -n base conda-libmamba-solver
54
-
conda config --set solver libmamba
53
+
conda config --set auto_activate_base false
55
54
56
-
We recommend to deactivate that conda self-activates its ``base`` environment.
57
-
This `avoids interference with the system and other package managers <https://collegeville.github.io/CW20/WorkshopResources/WhitePapers/huebl-working-with-multiple-pkg-mgrs.pdf>`__.
55
+
In order to make sure that the conda configuration uses ``conda-forge`` as the only channel, which will help avoid issues with blocked ``defaults`` or ``anaconda`` repositories, please set the following configurations:
58
56
59
57
.. code-block:: bash
60
58
61
-
conda config --set auto_activate_base false
59
+
conda config --add channels conda-forge
60
+
conda config --set channel_priority strict
62
61
63
62
.. code-block:: bash
64
63
65
-
conda create -n warpx -c conda-forge warpx
66
-
conda activate warpx
64
+
mamba create -n warpx -c conda-forge warpx
65
+
mamba activate warpx
67
66
68
67
.. note::
69
68
70
-
The ``warpx`` `conda package <https://anaconda.org/conda-forge/warpx>`__ does not yet provide GPU support.
69
+
The ``warpx`` package on conda-forge does not yet provide `GPU support <https://github.com/conda-forge/warpx-feedstock/issues/89>`__.
0 commit comments