[ENH]: generalize artists opting in-to auto limits · Issue #25139 · matplotlib/matplotlib · GitHub

Problem

This is triggered by #25127 most recently, but this is a discussion we have had a couple of times.

The core of the problem is that users have different expectations about what things will automatically adjust the axes limits (or not) in autolim mode. Currently Lines, Patches and AxesImage obligatorily participate

matplotlib/lib/matplotlib/axes/_base.py

Lines 2459 to 2483 in a046ee3

    
               def relim(self, visible_only=False): 
        
                   """ 
        
                   Recompute the data limits based on current artists. 
        
                   At present, `.Collection` instances are not supported. 
        
                   Parameters 
        
                   ---------- 
        
                   visible_only : bool, default: False 
        
                       Whether to exclude invisible artists. 
        
                   """ 
        
                   # Collections are deliberately not supported (yet); see 
        
                   # the TODO note in artists.py. 
        
                   self.dataLim.ignore(True) 
        
                   self.dataLim.set_points(mtransforms.Bbox.null().get_points()) 
        
                   self.ignore_existing_data_limits = True 
        
                   for artist in self._children: 
        
                       if not visible_only or artist.get_visible(): 
        
                           if isinstance(artist, mlines.Line2D): 
        
                               self._update_line_limits(artist) 
        
                           elif isinstance(artist, mpatches.Patch): 
        
                               self._update_patch_limits(artist) 
        
                           elif isinstance(artist, mimage.AxesImage): 
        
                               self._update_image_limits(artist)

and collections may optionally participate, but only when they are added

matplotlib/lib/matplotlib/axes/_base.py

Lines 2260 to 2273 in a046ee3

    
           if autolim: 
        
               # Make sure viewLim is not stale (mostly to match 
        
               # pre-lazy-autoscale behavior, which is not really better). 
        
               self._unstale_viewLim() 
        
               datalim = collection.get_datalim(self.transData) 
        
               points = datalim.get_points() 
        
               if not np.isinf(datalim.minpos).all(): 
        
                   # By definition, if minpos (minimum positive value) is set 
        
                   # (i.e., non-inf), then min(points) <= minpos <= max(points), 
        
                   # and minpos would be superfluous. However, we add minpos to 
        
                   # the call so that self.dataLim will update its own minpos. 
        
                   # This ensures that log scales see the correct minimum. 
        
                   points = np.concatenate([points, [datalim.minpos]]) 
        
               self.update_datalim(points)

Proposed solution

The proposed solution is to:

move the _update_line_limits and friends to the respective Artists
add an analogous method to the base Artist (probably defaulting to failure)
add a "I would like to particpate in autolimiting!" flag to base Artist
in relim look at the flag and call the newly generalized method above on any artists that opt-in

This will involved a little bit of public API (how to set the state to opt-in) and a bunch of private API (what should the signature of the method be).

A major difficulty is that many of our Artists may be anchored in data co-ordinates, but their bounding boxes are not in data-coordinates.

True, but now we can stuff that complexity back to the Artist classes where we have the best chance of knowing how to actually resolve that problem.

See #15595, which I had tried to push for during a sprint.

oh, nice that even has some good discussion over the tricky parts of the API (that I participated in and have no memory of 😞 )!

[ENH]: generalize artists opting in-to auto limits · Issue #25139 · matplotlib/matplotlib · GitHub

Repository navigation

[ENH]: generalize artists opting in-to auto limits #25139

Problem

Proposed solution

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

	def relim(self, visible_only=False):
	"""
	Recompute the data limits based on current artists.

	At present, `.Collection` instances are not supported.

	Parameters
	----------
	visible_only : bool, default: False
	Whether to exclude invisible artists.
	"""
	# Collections are deliberately not supported (yet); see
	# the TODO note in artists.py.
	self.dataLim.ignore(True)
	self.dataLim.set_points(mtransforms.Bbox.null().get_points())
	self.ignore_existing_data_limits = True

	for artist in self._children:
	if not visible_only or artist.get_visible():
	if isinstance(artist, mlines.Line2D):
	self._update_line_limits(artist)
	elif isinstance(artist, mpatches.Patch):
	self._update_patch_limits(artist)
	elif isinstance(artist, mimage.AxesImage):
	self._update_image_limits(artist)

	if autolim:
	# Make sure viewLim is not stale (mostly to match
	# pre-lazy-autoscale behavior, which is not really better).
	self._unstale_viewLim()
	datalim = collection.get_datalim(self.transData)
	points = datalim.get_points()
	if not np.isinf(datalim.minpos).all():
	# By definition, if minpos (minimum positive value) is set
	# (i.e., non-inf), then min(points) <= minpos <= max(points),
	# and minpos would be superfluous. However, we add minpos to
	# the call so that self.dataLim will update its own minpos.
	# This ensures that log scales see the correct minimum.
	points = np.concatenate([points, [datalim.minpos]])
	self.update_datalim(points)

[ENH]: generalize artists opting in-to auto limits · Issue #25139 · matplotlib/matplotlib · GitHub

Repository navigation

[ENH]: generalize artists opting in-to auto limits #25139

Description

Problem

Proposed solution

Activity

jklymak commented on Feb 2, 2023

tacaswell commented on Feb 2, 2023

anntzer commented on Feb 2, 2023

tacaswell commented on Feb 2, 2023

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions