added cyclepoint grouping with subgraphs

[markgrahamdawson]

Partly addresses #1130

• Added button to graph view to group nodes by cycle point.
• Works for timeseries and integer
• Option state saves when moving between workflows

---initial notes on work---
For the time being - figure out a means of drawing a dotted box around nodes that share a cyclepoint.
This can be done with graphviz subgraphs
The graphviz dot code is returned from the getDotCode function in Graph.vue view
Dot code can be rendered with
dot foo.dot -Tpng -o foo.png
ultimately in the svg there will need to be a rect tag added...

Check List

• I have read CONTRIBUTING.md and added my name as a Code Contributor.
• Contains logically grouped changes (else tidy your branch by rebase).
• Does not contain off-topic changes (use other PRs for other changes).
• Applied any dependency changes to both setup.cfg (and conda-environment.yml if present).
• Tests are included (or explain why tests are not needed).
• CHANGES.md entry included if this is a change that can affect users
• Cylc-Doc pull request opened if required at cylc/cylc-doc/pull/XXXX.
• If this is a bug fix, PR should be raised against the relevant ?.?.x branch.

[markgrahamdawson]

The amount of margin between subgraphs is random unless made explicit

No margin

Margin set in code

Im guessing the second one is better?

[hjoliver]

Def prefer the second one.

[MetRonnie]

Looking good, partial review so far

[oliver-sanders]

Looks good, works nicely 👍

Sadly, GraphViz doesn't stop edges from being routed through graph labels by the looks of it. Not much we can do about that, but we could try adding a white stroke (SVG lingo for border) around the text to keep it readable, e.g:

element {
    fill: black;
    stroke-width: 2;
    stroke: white;
}

Could do with bumping the font-size of the graph labels a bit, not much, they don't need to be larger than task GraphNodes, but they do get lost in big graphs rather easily e.g:

[oliver-sanders]

Bumping this forward to the 8.3 meta-release as this should be ready to go in before cylc-flow blockers.

It would be good to get a screenshot of this into the "changes" section in cylc-doc: https://cylc.github.io/cylc-doc/nightly_8.3/html/reference/changes.html#cylc-8-3

[MetRonnie]

Got a PR against this at markgrahamdawson#5

[MetRonnie]

Managed to get a workflow into a state where there's an grouping with the title "undefined".

And transposing this graph gives a console error (resolved)

[markgrahamdawson]

to a state where there's an grouping with the title "undefined". And transposing this graph gives a console error. Will see if I can figure out anything else tomorrow

Do you have the flow.cylc file?

[oliver-sanders]

Hmm, I didn't notice any issues with the complex workflow when I tested it. Will need the dot code to debug this. Given that the edges (yes that's right word) lead outside of the subgraph, I expect there may be an issue with the formatting of the subgraph dot code.

When I read through the code I think I spotted a missing newline at the end of the subgraph. I thought it was harmless, but it might be worth checking that's not causing issues.

[MetRonnie]

Managed to get a workflow into a state where there's an grouping with the title "undefined"

Also the 3 tasks are showing as "state unknown" while the jobs are showing as running. But these are from an earlier cycle and have actually already succeeded. Might be something to do with simulation mode. Or actually #1635?

[MetRonnie]

Got a fix for the arrows pointing nowhere (and the console error) in markgrahamdawson#5

[oliver-sanders]

Regards that "undefined" error for tasks with the missing cyclePoint field.

There is a slightly nicer way to fetch the cycle point that is available for all tasks by default (no need to request the cyclePoint field explicitly) using the "tokens" which are provided for every object in the store. Due to the way tokens are provisioned, the cycle cannot be undefined by accident (it's derived from the ID).

I haven't tested with these changes so worth giving them a spin before committing to them, but hopefully that'll resolve the error.

[oliver-sanders]

(re-approving)

[oliver-sanders]

👍

[oliver-sanders]

(added a changes entry in cylc-doc cylc/cylc-doc#733)