Enh control tick deconflict2 #13908
Enh control tick deconflict2 #13908
Conversation
| @@ -754,6 +756,17 @@ def __init__(self, axes, pickradius=15): | |||
| majorTicks = _LazyTickList(major=True) | |||
| minorTicks = _LazyTickList(major=False) | |||
|
|
|||
| def get_remove_overlapping_locs(self): | |||
There was a problem hiding this comment.
I thought we decided not to bother with getter and setters?
There was a problem hiding this comment.
I currently cannot think of a reason not to use the flag on the axis instead of the locator. In turn, a clear benefit is that major locators do not need to carry a useless (confusing) attribute with them. So this solution seems good to me.
I could imagine that if people are using this they might want to have an rc parameter for it, but one can sure wait for user feedback first.
Concerning getters/setters, maybe it's beneficial to have something like
ax.xaxis.set(minor_locator=MultipleLocator(30), remove_overlapping_locs=False)
working as expected? This would require there to be a set_remove_overlapping_locs method, right?
|
I indeed missed the fact that this now lives on an Artist subclass and therefore the setter is useful to play nicely with Artist.set, so I'm fine with keeping it. |
|
This looks fine to me. Definitely needs documentation somewhere... |
|
API changes notes are pushed to #13870 I don't think this needs a more prominent tutorial or example. |
|
(specifically 6cbbc85 ) |
|
Not clear what is wrong with appeyor... |
|
rebase and push? |
In matplotlib#13363 when `iter_ticks` was deprecated the in-lined logic did not account for the updates from matplotlib#13314.
If Axis.remove_overlapping_locs is set to False the `Axis` object will not attempt to remove locations in the minor ticker that overlap with locations in the major ticker. This is a follow up to matplotlib#13314 which un-conditionally removed the overlap. closes matplotlib#13618
This will make sure extra ticks (that will not be used) will not be returned. The change to `get_major_ticks` is a no-op, but done for symmetry.
This is to prevent user confusion if they are reaching directly into `Axis.minorTicks` and `Axis.majorTicks` and see ticks report they are visible but are not actually drawn.
0ed8a8c to
0a6f575
Compare
|
rebased on current master and force-pushed, appveyor looks happier (it started!). |
OK, but so far as I can see this has no documentation at all except for the API note? https://20198-1385122-gh.circle-artifacts.com/0/home/circleci/project/doc/build/html/api/axis_api.html#axis-objects |
|
Pushed a docs change. |
…908-on-v3.1.x Backport PR #13908 on branch v3.1.x (Enh control tick deconflict2)
This reverts part of the changes from matplotlib#13908 (in particular commit e908efe ) that tried to manage the visibility if the ticks that are not used. However, there was a bug in this that nothing would ever set ticks to be visible again. Because we currently cache / recycle ticks this meant that panning / zooming or resizing the figure would eventually lose ticks if the number of ticks ever went down and then back up. We could fix this by adding a `tick.set_visible(True)` for the drawn ticks, however this would break the (brittle) use case where users are reaching in and manually setting the visibility of individual ticks to False to hide them. Instead, we revert the consistency code altogether and accept that there will be `Tick` objects in `Axis.majorTicks` and `Axis.minorTicks` that are marked as visible, but are not drawn (because they are filtered by `_update_ticks`).
This reverts part of the changes from matplotlib#13908 (in particular commit e908efe ) that tried to manage the visibility if the ticks that are not used. However, there was a bug in this that nothing would ever set ticks to be visible again. Because we currently cache / recycle ticks this meant that panning / zooming or resizing the figure would eventually lose ticks if the number of ticks ever went down and then back up. We could fix this by adding a `tick.set_visible(True)` for the drawn ticks, however this would break the (brittle) use case where users are reaching in and manually setting the visibility of individual ticks to False to hide them. Instead, we revert the consistency code altogether and accept that there will be `Tick` objects in `Axis.majorTicks` and `Axis.minorTicks` that are marked as visible, but are not drawn (because they are filtered by `_update_ticks`). Closes matplotlib#14054
This is an alternative to #13826 that puts the knob on the
Axisrather than the locator.I have a memory of talking about doing this and deciding not to, but do not remember why.
attn @efiring