You are here

Switch to conda-forge for installing data reduction software

Content owned by klabrie

STScI has recently announced that it will end support for the main Astroconda channel, used for installing Gemini data reduction packages, on 1 February 2023.

We were already planning to base our upcoming DRAGONS 3.1 release on the more up-to-date package set at conda-forge and have been working for the past few weeks to ensure that dependencies of Gemini IRAF and DRAGONS 3.0 can also be installed using conda-forge, in place of STScI's Astroconda channel and Anaconda's "defaults". As of writing, these instructions also apply to the latest (July 2022) version of the Gemini IRAF VM.

All users are advised to remove STScI's channel from your Anaconda configuration as follows:

  conda config --remove channels

Until this is done, you may find it impossible to install or update any conda packages at all after 1 February.

You should also add conda-forge and Gemini's public channel as follows -- even if you already have them defined (to make sure they appear in the correct order of precedence):

  conda config --add channels conda-forge
  conda config --add channels

When you next need to update packages, instead of doing so within your existing ("geminiconda" or similar) environment, you should create a new environment, to get a consistent set of packages from the new channels; Gemini will not be able to troubleshoot problems due to installation from an unsupported mixture of channels.

An environment with DRAGONS and Gemini IRAF can be created in much the same way as previously, omitting any "--no-channel-priority" flag. You will no longer be able to install the "stsci" meta-package alongside our software, however, since STScI is dropping support for it. If you need both Gemini and STScI data reduction packages, you should create separate conda environments for each one and follow STScI's instructions at the above link, to set up "stenv". Since DS9 was previously part of "stsci", you should now include that package separately. You should also specify "dragons" explicitly, to help ensure that the latest version gets picked up.

In summary, you can type the following on Linux:

  conda create -n geminiconda3 python=3.7 iraf-all pyraf-all ds9 dragons

(where "geminiconda3" is whatever you decide to call the new environment, to distinguish it from the old one).

On MacOS 10.15+, you should omit the IRAF-related packages, since they can no longer run there natively, without the IRAF (Linux) VM:

  conda create -n dragons python=3.7 ds9 dragons

Anyone still wanting to run PyRAF natively on MacOS 10.13-10.14 should specify "python=3.7.9" to get a suitable combination of packages on those old OSs, as a workaround for several conda-forge and conda limitations.

Because the older Python 2 stack is heavily dependent on Astroconda, it may no longer be possible to maintain Python 2 environments after January, but you may continue using any that you already have, until something needs updating; all new environments should use Python 3 (see our announcement from 23 August 2022).

We appreciate your patience while any remaining wrinkles are ironed out; any further updates will be announced here.

Data Reduction Software
News Category
Science Operations Announcements

News Archive Filter

Switch to conda-forge for installing data reduction software | Gemini Observatory


The website encountered an unexpected error. Please try again later.