GeezerVison

Marking as draft for now due to failing Jasmine test (https://github.com/plotly/plotly.js/actions/runs/27172958527/job/80215928449?pr=7830)

I changed the base branch to v4.0 since this will be a breaking change.

Let me know if you need help with the tests. Those sankey ones are pretty brittle.

One last point that I found while testing things:

In @plotly/[email protected], calling sankey.nodePadding() after running the layout returned the effective (post-clamp) padding value. In 0.12.3 the getter returns the user-configured value instead, because upstream split the
internal state into separate dy (configured) and py (effective) variables.

This silently breaks the warning at src/traces/sankey/render.js#L76-L78:

if (sankey.nodePadding() < nodePad) {
    Lib.warn('node.pad was reduced to ', sankey.nodePadding(), ' to fit within the figure.');
}

After the upgrade, sankey.nodePadding() always equals nodePad, so the comparison is never true and the warning never fires. The layout itself still clamps correctly - only the diagnostic is affected.

Options to address the issue:

• Remove the warning entirely
• Derive the effective padding from the laid-out node positions (e.g. measure the gap between consecutive nodes in the densest column)

Options to address the issue:

Remove the warning entirely
Derive the effective padding from the laid-out node positions (e.g. measure the gap between consecutive nodes in the densest column)

Deriving the padding sounds fairly complicated just for the sake of a warning, but maybe we could expose the effective node padding with a new method in the sankey library to keep this behaviour.

I'll file an issue and we can punt for now. It's not critical.

Thanks for fixing the tests! Sorry I've been busy lately and hadn't got around to this

No worries. That's collaboration!