Add automatic categorical legend for datashaded plots

[philippjfr]

Inspects the pipeline of an RGB element and discovers the shade and rasterize operations used to generate an RGB to then automatically add a legend.

• Clean up operation and factor out operations to traverse pipeline
• Factor out code to render legend plot and use across all interfaces
• Update legend when needed

[jbednar]

Does this still have the property that the previous hack did, of disrupting the axis range of a plot if that plot didn't happen to include 0,0? If so that needs documenting and explaining how to work around it.

[philippjfr]

Does this still have the property that the previous hack did, of disrupting the axis range of a plot if that plot didn't happen to include 0,0? If so that needs documenting and explaining how to work around it.

No, apply_ranges=False should avoid this.

[philippjfr]

Actually no, not even that, it won't cause that issue because the plot won't even take part in axis range calculation in this approach.

[jbednar]

Woohoo!!!

[philippjfr]

Okay, happier with this now. I've put the operation behind a try/except as well so we don't cause any regressions in case I missed something.

[jbednar]

How can I get text rather than numeric labels for a plot like this?

Also, I'm getting warnings simply from importing HoloViews from this branch, with param 1.10.1 and pandas 1.1.5 installed:

[philippjfr]

Also, I'm getting warnings simply from importing HoloViews from this branch, with param 1.10.1 and pandas 1.1.5 installed:

Already fixed in #4821

How can I get text rather than numeric labels for a plot like this?

Hmm, I suppose that's a problem. It uses what's actually in the data (which is how legends everywhere work). We could introduce a legend_labels option to allow remapping labels I suppose.

[codecov-commenter]

Codecov Report

Merging #4806 (8bcd688) into master (c265cc1) will decrease coverage by 0.09%.
The diff coverage is 53.29%.

@@            Coverage Diff             @@
##           master    #4806      +/-   ##
==========================================
- Coverage   86.74%   86.64%   -0.10%     
==========================================
  Files         299      299              
  Lines       62218    62343     +125     
==========================================
+ Hits        53969    54019      +50     
- Misses       8249     8324      +75     

Impacted Files	Coverage Δ
holoviews/plotting/plot.py	93.01% <ø> (ø)
holoviews/plotting/mpl/raster.py	78.52% <31.57%> (-3.51%)	⬇️
holoviews/plotting/util.py	69.49% <32.35%> (-2.06%)	⬇️
holoviews/operation/datashader.py	78.33% <43.39%> (-1.72%)	⬇️
holoviews/plotting/bokeh/raster.py	71.30% <65.21%> (-0.95%)	⬇️
holoviews/operation/element.py	69.90% <66.66%> (-0.06%)	⬇️
holoviews/plotting/mpl/element.py	89.94% <96.15%> (-0.09%)	⬇️
holoviews/plotting/bokeh/element.py	88.86% <100.00%> (ø)
holoviews/plotting/mpl/chart.py	64.02% <100.00%> (ø)
... and 1 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c265cc1...8bcd688. Read the comment docs.

[bbudescu]

Can this be used for showing legend entries of rasterized hv.Curves, too?

[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.