Autoranging on multi-line plots limits view to a single line

[droumis]

ALL software version info

Python : 3.10.13 | packaged by conda-forge | (main, Oct 26 2023, 18:09:17) [Clang 16.0.6 ]
Operating system : macOS-13.5.2-arm64-arm-64bit
Panel comms : default

holoviews : 1.18.0
bokeh : 3.3.0
hvplot : 0.9.0
jupyterlab : 4.0.7

Description of expected behavior and the observed behavior

Interacting, via zoom or pan, on a plot with multiple lines limits the viewport to a single line if autorange='y' is set.

Complete, minimal, self-contained example code that reproduces the issue

import pandas as pd
import numpy as np
import hvplot.pandas

length = 100
parameters = [
    (0, 1),      # Mean 0, SD 1
    (100, 20),   # Mean 100, SD 20
    (1000, 200), # Mean 1000, SD 200
    (-500, 50)   # Mean -500, SD 50
]

df_ = pd.DataFrame({f"{i}": np.random.normal(mean, sd, length) for i, (mean, sd) in enumerate(parameters)})
df_ = df_.reset_index().melt(id_vars=['index'], var_name='source', value_name='value')
df_['source'] = df_['source'].astype(int)

# Single Line Plot
s = df_.hvplot(x="index", y="value", downsample=True, autorange='y').opts(width=450, shared_axes=False)

# Multi Line Plot
m= df_.hvplot(x="index", y="value", by='source', downsample=True, autorange='y').opts(show_legend=False, width=450, shared_axes=False)

s + m

Stack traceback and/or browser JavaScript console output

Screenshots or screencasts of the bug in action

Screen.Recording.2023-11-04.at.8.52.35.AM.mov

• I may be interested in making a pull request to address this

[droumis]

And autorange=y does not seem to work at all with multiple lines using just HoloViews:

import holoviews as hv; hv.extension('bokeh')
import numpy as np

np.random.seed(42)
ys = np.random.randn(101).cumsum(axis=0)

curve = hv.Curve(ys) * hv.Curve(ys[::-1])
curve.opts(autorange='y')

Screen.Recording.2023-12-14.at.2.11.23.PM.mov

[droumis]

Update from @jlstevens:

Together with @maximlt, they found out there is a more general problem with autoranging and overlays that is probably the cause; probably a regression since the multi-axis work. As such it will need to be fixed for the next holoviews release.

@jlstevens will have another look this week

[droumis]

@jlstevens, any update from last week?

[jlstevens]

I'm currently working on it - unfortunately, this is a really tricky bit of code!

[droumis]

Hi @jlstevens, could you provide a status update on this?

[jlstevens]

I had to step away from my debugging attempts for a while but I can now look into this once more. My goal is to open a PR this week. Thank you for the reminder!

[droumis]

Hi @jlstevens, any progress on that PR?

[jlstevens]

Sorry I got distracted by other priorities once again. Perhaps there is someone else who could be assigned to work together with me on this? That will help keep me on track and help get this fixed faster.

[philippjfr]

Turned out to be a small oversight in the existing implementation, so I fixed it.

[jlstevens]

Thank you!

I really thought I had to do something in the Python code to accumulate over the elements of the overlay. I realize my mental model of how the callback was looping over elements was wrong.

Anyway, I'm just glad it is fixed :-)

[github-actions]

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