Repository navigation
sticky_edges behavior is not consistent within/across subpots #15967
Description
Bug report
Bug summary
I am trying to draw lines that do not get a margin placed around them when the axis is autoscaled. My understanding is that the current way to accomplish that is with Artist.sticky_edges. But I am finding that while this works within a single Axes, it does not work across multiple Axes
Code for reproduction
This works:
f, ax = plt.subplots()
ax.scatter([0, 1], [0, 0])
x1 = ax.get_xlim()
line, = ax.plot(x1, [0, 0])
line.sticky_edges.x[:] = x1
ax.scatter([0, 1], [1, 1])
x2 = ax.get_xlim()
line, = ax.plot(x2, [1, 1])
line.sticky_edges.x[:] = x2
assert x1 == x2While this fails:
f, (ax1, ax2) = plt.subplots(1, 2, sharex=True, sharey=True)
ax1.scatter([0, 1], [0, 0])
x1 = ax1.get_xlim()
line, = ax1.plot(x1, [0, 0])
line.sticky_edges.x[:] = x1
ax2.scatter([0, 1], [1, 1])
x2 = ax2.get_xlim()
line, = ax2.plot(x2, [1, 1])
line.sticky_edges.x[:] = x2
assert x1 == x2Is this expected to work? I know there are active changes to the autoscaling behavior; is this a bug that has been fixed (I'm not in a position to test things on a development build right now). Is there a different way to accomplish "ignore an artist when autoscaling"?
Matplotlib version
- Operating system: macos
- Matplotlib version: 3.1.1
- Matplotlib backend (print(matplotlib.get_backend())): inline
- Python version: 3.7
Activity
anntzer commented on Dec 17, 2019
Likely #15595 (pretty close to merging I think, just need to allocate some time for it...) will help?
sticky_edges across shared axes should work, but bugs may be present for sure.
mwaskom commented on Dec 17, 2019
That method does look useful, although I like how sticky_edges has axis-specificity, which set_in_autoscale appears not to have?
anntzer commented on Dec 17, 2019
Fair point, I guess set_in_autoscale could allow ("relatively" easily) controlling x/y separately, do you want to leave a note in #15595 to that effect?
mwaskom commented on May 6, 2021
FYI this seems to have been fixed at some point. I'm going to close, but if someone knows why/when it was fixed that would be helpful, because it just caused my pinned CI build to fail.
Skip test that fails because of matplotlib bug
QuLogic commented on May 6, 2021
The original example was fixed in 167105f. #16450, and 3.2.0.
Several fixes/improvements related to lmplot axis scaling (#2576)
Several fixes/improvements related to lmplot axis scaling (#2576)