From d4301a1ad4ea652e2c5c3e0dd09fa4c888881731 Mon Sep 17 00:00:00 2001 From: Victor Lin <13424970+victorlin@users.noreply.github.com> Date: Tue, 27 May 2025 14:42:40 -0700 Subject: [PATCH 1/2] Remove warning on using Apple silicon version of Conda Improvements have been made by the Bioconda community to support more osx-arm64-native packages. I've tested that the example ambient setup command now works without any problems using an Apple silicon version of Micromamba. --- src/install.rst | 4 ---- 1 file changed, 4 deletions(-) diff --git a/src/install.rst b/src/install.rst index 08587318..963c73c3 100644 --- a/src/install.rst +++ b/src/install.rst @@ -225,10 +225,6 @@ Set up a Nextstrain runtime .. group-tab:: macOS - .. warning:: - - If step 2 fails, you might have an Apple silicon version of Conda installed. See :ref:`this FAQ section ` for workarounds. - .. include:: snippets/ambient-setup.rst From 67b39a1134c36a1f24650ff6949465ea2a409c74 Mon Sep 17 00:00:00 2001 From: Victor Lin <13424970+victorlin@users.noreply.github.com> Date: Tue, 27 May 2025 15:05:13 -0700 Subject: [PATCH 2/2] Remove recommendation to use Intel-based Conda installation Replaced by another FAQ entry that describes the previous recommendation and why it was removed. --- src/reference/faq.rst | 49 ++++++------------------------------------- 1 file changed, 6 insertions(+), 43 deletions(-) diff --git a/src/reference/faq.rst b/src/reference/faq.rst index 03d5cf14..51b29ca3 100644 --- a/src/reference/faq.rst +++ b/src/reference/faq.rst @@ -89,56 +89,19 @@ The other deciding factor is which runtime(s) you'd like to use. Under **PowerSh .. old anchors .. _why-intel-miniconda-installer-on-apple-silicon: - .. _why-conda-install-errors-on-apple-silicon: -Why do I get an error when installing to my Conda environment on a Mac computer with Apple silicon (e.g. M1)? -------------------------------------------------------------------------------------------------------------- - -An example error: - -.. code-block:: text - :emphasize-lines: 10,11,12,13,14,15 - - (your-environment-name) $ conda install augur - - … - - Platform: osx-arm64 - - … - - Collecting package metadata (repodata.json): done - Solving environment: failed - - LibMambaUnsatisfiableError: Encountered problems while solving: - - nothing provides fasttree needed by augur-10.0.0-py_0 - - Could not solve for environment specs - -This happens when using an ARM64-based Conda installation on a `computer with Apple silicon `__, but there are workarounds. - -Apple silicon chips are great and efficient. They are based on a different chip architecture, ARM64/AArch64, and come with performance improvements compared to the x64-based Intel chips in older Macs. - -However, many existing packages have not yet added support to run on these chips natively. An easy way to identify support on the `Bioconda packages page `_ is to look for ``noarch`` or ``osx-arm64`` under the **Installers** section of a package. Without any of those, a package is not able to be installed natively on Apple silicon. This is the case for packages such as `MAFFT `_ (a dependency of :term:`Augur`) and many other bioinformatics packages. For this reason, using an ARM64-based Conda installation for the average bioinformatics researcher can result in a difficult experience. - -There are two ways to work around this: - -1. Uninstall Conda, delete all existing environments, and re-install with an `Intel-based installer `__. With an Intel-based installation, all environments are forced to use emulation. - - This provides easy compatibility with a broader set of bioinformatics packages, but comes at the cost of relatively longer run times for packages that have native ARM64 support. - -2. Create a custom Conda environment that installs and runs packages under `Intel emulation `__. Run this after setting up an **empty** Conda environment and before installing any packages to it: +.. _what-happened-to-the-intel-conda-recommendation: - .. code-block:: bash +What happened to the recommendation to use an Intel-based Conda installation? +----------------------------------------------------------------------------- - conda config --env --set subdir osx-64 +Background: computers with Apple silicon hardware have two options for Conda subdir: ``osx-arm64`` (native) and ``osx-64`` (Intel-based, requires Rosetta 2 emulation). - This will ensure that all commands in the active Conda environment are run using Intel emulation, making it possible to install Nextstrain software such as Augur. You only need to run this once per Conda environment. +When Apple silicon was first released, few Conda packages supported ``osx-arm64``. This meant that the example ``conda create`` command provided in the Nextstrain docs would fail to resolve packages when using an ``osx-arm64`` Conda installation. We had suggested using an ``osx-64`` Conda installation as an easy workaround for most users, at the cost of not taking full advantage of the Apple silicon architecture. - .. warning:: +Now that there is wide support for ``osx-arm64`` packages, we have removed the workaround. - This should only be done on an empty Conda environment (otherwise you may encounter low-level errors) and does not automatically apply to other new or existing environments. .. _what-happened-to-the-native-runtime: