Apply autorange on data change events

[philippjfr]

• Had to schedule the CustomJS callback using setTimeout because the ColumnDataSource.data callback fires before the glyphs (and more importantly their indexes) are updated.
• We remove all usage of cb_obj in the callback which allows us to reuse the callback between both data and range updaates.
• We add the autorange callback to a new _js_on_data_callbacks instance variable which then gets applied to all sources attached to a GlyphRenderer

ToDo

• Check inverted axes work with this
• Guard against failed lookups in Bokeh.index
• Guard the view.child_views lookup to avoid errors in layouts
• Somehow filter out GlyphRenderers which were generated by plots with apply_ranges=False (suggest using .tags property to mark them)

[codecov-commenter]

Codecov Report

Merging #5609 (7f0a876) into main (a441bfa) will decrease coverage by 0.03%.
The diff coverage is 44.44%.

@@            Coverage Diff             @@
##             main    #5609      +/-   ##
==========================================
- Coverage   88.29%   88.27%   -0.03%     
==========================================
  Files         302      302              
  Lines       62460    62495      +35     
==========================================
+ Hits        55152    55168      +16     
- Misses       7308     7327      +19     

Impacted Files	Coverage Δ
holoviews/plotting/bokeh/element.py	88.17% <44.44%> (-0.44%)	⬇️
holoviews/core/accessors.py	83.72% <0.00%> (-1.66%)	⬇️
...loviews/tests/plotting/matplotlib/test_renderer.py	97.50% <0.00%> (-1.21%)	⬇️
holoviews/operation/datashader.py	83.90% <0.00%> (-0.09%)	⬇️
holoviews/tests/core/test_dynamic.py	99.45% <0.00%> (+<0.01%)	⬆️
holoviews/tests/plotting/bokeh/test_callbacks.py	98.65% <0.00%> (+<0.01%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[jlstevens]

I think holoviews is doing what I asked for here even though the behavior is a little crazy!

[jbednar]

I think holoviews is doing what I asked for here even though the behavior is a little crazy!

That doesn't seem right at all. inverting should be about how that particular plot is displayed, not any other plot...

[philippjfr]

That doesn't seem right at all. inverting should be about how that particular plot is displayed, not any other plot...

Linked axes are shared axes, so there's no way to avoid this.

[jbednar]

Linked axes are shared axes, so there's no way to avoid this.

I'd say that there's no current way. :-) I'd consider this a bug, since I think the intent of the user is clear here: share the range extents, don't share the orientation. Addressing it seems very difficult, so maybe there can be a warning if axes have different values for inversion, saying that invert_{xy}axis must be set to the same value on all linked axes (so that users realize they need to set it on both or neither, with the current implementation).

[jlstevens]

I think a warning for now and an issue to record the eventual desired behavior is the way forward for now...

[jlstevens]

@philippjfr I've finished off the tasks mentioned in your TODO list - the tags suggestion works great! Ready for review/merge once the test go green.

[philippjfr]

Happy to see this merged. Since this is my PR I can't approve, so feel free to approve yourself and merge.

[jlstevens]

I'll add a line to the user guide about auto ranging applying after data updates and then I'll merge.

[jlstevens]

I'll go ahead and merge.

[github-actions]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.