This update upgrades cylc-uiserver to work with the most recent versions of Jupyter components, namely:
Jupyter Server v2.7+
Jupyter Hub v4.0+
This release offers some big improvements for sites looking to deploy the Cylc GUI with Jupyter Hub:
Users can now be authorised to start UI Servers on behalf of other users.
There is now a box in the top-left of the GUI for switching user / deployment.
The Cylc UI Server can now be deployed along with the latest versions of Jupyter Lab.
Upgrading an existing Jupyter Hub deployment to use cylc-uiserver requires some extra configuration due to the new authorisation frameworks in Jupyter Hub and Jupyter Server. For more information on how to set up a Jupyter Hub deployment with cylc-uiserver 1.4.0 see this page:
This release contains bugfixes for playing workflows. See the changelog for details.
Note: If you installed version 1.4.2, please update to 1.4.3 promptly as 1.4.2 unfortunately contained a bug that broke the ability to play workflows (depending on your Cylc environment name).
This release also comes with UI v2.3.0 which fixes a bug in filtering workflows in the sidebar. See the changelog for details.
How to update
# For Conda managed environments:
conda update 'cylc-uiserver=1.4.3'
# For Pip managed environments:
pip install 'cylc-uiserver==1.4.*'
Rose now ignores PYTHONPATH to make it more robust to task environments which set this value. If you want to add to the Rose environment itself, e.g. to write a rose-ana test, use ROSE_PYTHONPATH.
cylc-flow 8.2.5 contains several bugfixes. See the changelog for details.
cylc-uiserver 1.4.4 contains an important bugfix for an issue that limited the number of active workflows that could be tracked at the same time. See the changelog for details.
cylc-ui 2.4.0 contains several enhancements and bugfixes:
Workspace tab layout is now remembered and restored when navigating between workflows (in
a single browser session; we are investigating saving layouts across browser sessions).
Does that mean you can’t use -S or -O or -D in a vr, and only in vip for example? If so, if you have a running workflow, and you want to update a jinja variable passed in via -S, how do you do it? I’m guessing vr still can reference -S, etc as I think the r is reinstall?
Yes, the -S, -O and -D options still work as intended with cylc vip and cylc vr, so you should be able to update a Jinja2 variable like so cylc vr <id> -S KEY=<value>. These options apply at install time i.e. cylc install (the “i” in “vip”) and cylc reinstall (the “r” in “vr”). They weren’t intended to be added to the cylc play command where they don’t really make sense.
This fixes an issue introduced in 8.2.6 (our apologies) which caused the Rose options -S, -O and -D to cause the cylc play command to fail for site configurations where Cylc is configured to detach the scheduler process on a remote host.
How to update site installations
Sites should create a new Cylc environment and manage parallel installations using the provided wrapper script.
How to update stand-alone installations
# First shut down running workflows:
cylc stop '*'
# For Conda managed environments:
conda update 'cylc-flow=8.2.7' 'cylc-rose=1.3.4'
# For Pip managed environments:
pip install 'cylc-flow==8.2.7' 'cylc-rose==1.3.4'