FIX: ignore axes that aren't visible

[jklymak]

PR Summary

Closes #16173

Layering subplots causes the topmost subplot to capture mouse events, which seems fine. However, if the topmost subplot is made invisible (which is a trick we use in cosntrained_layout) then it would still capture the mouse event. Given that an axes with ax.get_visble() is False is never drawn, not are any of its children, it seem unlikely that anyone would want to pick on a placeholder like that so the son here is to ignore invisible axes.

Note that I also changed how constrained_layout makes the axes invisible. It passes all the tests, so this is preferable to the way I was doing it before.

Test

import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
from matplotlib.widgets import Button

def on_click(event):
    print("hey")

fig = plt.figure(constrained_layout=True)

gs = gridspec.GridSpec(2, 1, figure=fig)

ax1 = fig.add_subplot(gs[0, 0])
ax1.text(0.5, 0.5, "ax1", va="center", ha="center")

gs2 = gs[1, 0].subgridspec(1, 2)
ax2 = fig.add_subplot(gs2[0, 1])

b = Button(ax2, 'hey')
b.on_clicked(on_click)

plt.show()

PR Checklist

• Has Pytest style unit tests
• Code is Flake 8 compliant
• New features are documented, with examples if plot related
• Documentation is sphinx and numpydoc compliant
• Added an entry to doc/users/next_whats_new/ if major new feature (follow instructions in README.rst there)
• Documented in doc/api/api_changes.rst if API changed in a backward-incompatible way

[anntzer]

I'll trust you that it fixes the issue reported on SO, but this certainly makes sense.

[jklymak]

Is there an easy way to test button clicks?

[anntzer]

It's slightly a pain, see

matplotlib/lib/matplotlib/tests/test_widgets.py

Line 20 in 860803d

def do_event(tool, etype, button=1, xdata=0, ydata=0, key=None, step=1):

or https://github.com/anntzer/mplcursors/blob/1ac466c613613edc8660161a407380db52985b48/tests/test_mplcursors.py#L59

[jklymak]

added test in test_axes.py

[ImportanceOfBeingErnest]

Due to the fix from f215d09 this is fine.

Though in general, I think not having any extra axes in the game would be much better.
Would it be possible to remove the axes after it has been used as placeholder?

[jklymak]

Not easily. The layout boxes used by constrained layout are attributes of the axes.

You could imagine re-architecting the whole thing so the layout box was the attribute of some sort of subplotspec object, but we don't have those (the axes also keeps track of its subplotspec). I don't think its too bad a hack, but certainly if we re-architect all the layout engines it should be considered.

[ImportanceOfBeingErnest]

I could imagine other problems occuring because of extra axes being present in the figure.
Maybe in the longer run one could imagine constrained_layout keeping track of the axes it needs, without them necessarily being part of the figure?

In any case, this fixes the bug, so approving as is.

[anntzer]

perhaps reverse the order of the checks? get_visible() should be faster (just an attribute check) than contains_point()
@jklymak can self-merge with or without this change.

[jklymak]

The first check is to make sure the pick is actually in the axes (because its in pixels). So if the second check failed, and you didn't have the first, you wouldn't know if it failed because of a resolution change in the tests or because the functionality actually broke.

In normal conditions both tests should pass, so I don't think there is a typical time savings here.

[lumberbot-app]

Owee, I'm MrMeeseeks, Look at me.

There seem to be a conflict, please backport manually. Here are approximate instructions:

1. Checkout backport branch and update it.

$ git checkout v3.2.x
$ git pull

2. Cherry pick the first parent branch of the this PR on top of the older branch:

$ git cherry-pick -m1 b6d333642e2cf8882183ca8071ad411ae5b19b73

3. You will likely have some merge/cherry-pick conflict here, fix them and commit:

$ git commit -am "Backport PR #16175: FIX: ignore axes that aren't visible"

4. Push to a named branch :

git push YOURFORK v3.2.x:auto-backport-of-pr-16175-on-v3.2.x

5. Create a PR against branch v3.2.x, I would have named this PR:

"Backport PR #16175 on branch v3.2.x"

And apply the correct labels and milestones.

Congratulation you did some good work ! Hopefully your backport PR will be tested by the continuous integration and merged soon!

If these instruction are inaccurate, feel free to suggest an improvement.