Cylc consists of multiple components.
This Conda environment will provide you with the “recommended” installation of all components:
name: cylc-8.0 channels: - conda-forge dependencies: - cylc-flow =8.0 - cylc-uiserver - cylc-rose - metomi-rose
The following sections outline alternatives for finer control over installation.
The simplest Cylc installation consists only of Cylc Flow:
name: cylc-flow channels: - conda-forge dependencies: - cylc-flow =8.0
We suggest installing Cylc Flow at a “minor” version (e.g.
8.1.2) to pick up the latest “maintenance” release.
If you do not specify your Python version you will be provided with the most recent compatible one.
If you are installing Cylc on multiple machines across a network it is advisable to keep the Python versions consistent.
You can do this by adding Python into the environment at a particular version:
name: cylc-flow-with-python-3.7 channels: - conda-forge dependencies: - python =3.7 - cylc-flow =8.0
Cylc Flow provides a cut-down package containing only the bare essentials
This may be useful for installing onto job hosts where client-facing extra features are not of interest:
name: cylc-flow-base channels: - conda-forge dependencies: - cylc-flow-base =8.0
name: cylc-uiserver channels: - conda-forge dependencies: - cylc-flow =8.0 - cylc-uiserver
To exclude Jupyter Hub from the installation use
name: cylc-uiserver-without-jupyterhub channels: - conda-forge dependencies: - cylc-flow =8.0 - cylc-uiserver-base
name: cylc-uiserver-with-traefik-proxy channels: - conda-forge dependencies: - cylc-flow =8.0 - cylc-uiserver-hub-base - pip - pip: - jupyterhub-traefik-proxy
For working with Rose add
cylc-rose (the bridge
between Cylc & Rose):
name: cylc-rose channels: - conda-forge dependencies: - cylc-flow =8.0 - cylc-rose - metomi-rose
We will look at providing more installation options for Rose in the near future.
Adding Cylc To Your Conda Package
If you want to publish a package (e.g. to Conda Forge) that depends on Cylc,
consider using the minimal package (e.g.
cylc-flow-base) as a dependency
to allow the installer to maintain flexibility over the installation.
Working With Other Conda Channels
Cylc projects are published to Conda Forge and the above environments install all dependencies from Conda Forge.
If you want to install other dependencies (e.g. Python), from other channels (e.g. Anaconda), list the dependency explicitly and place the channel above the Conda Forge channel.
name: cylc-flow channels: - defaults - conda-forge dependencies: - python - cylc-flow =8.0