wrong x-position of marker with drawstyle='steps-xxx'

Bug report

Hello, as discussed on matplotlib-devel, there is a unexpected change in Matplotlib 2.2 when plotting lines and markers (e.g. style 'D-') combined with drawstyle='steps-post' (steps-pre' and 'steps-mid' as well).

Code for reproduction

import matplotlib.pyplot as plt
import numpy as np
x = np.linspace(0, 6, 7)
y = np.sin(x)
plt.plot(x, y, 'D-', drawstyle='steps-post')

Actual outcome

In Matplotlib 2.2.2, markers are drawn at each "line break".

Running the same script with drawstyle='steps-mid' yields markers at half-integer positions (except for those at the end)

Expected outcome

In all previous Matplotlib versions, the marker was placed at the exact x-position of the input data (which is the left of each segment for drawstyle='steps-post'). I didn't do the bissection to know when the changed happen, but I suspect it's quite new (2.1 → 2.2 ?).

Matplotlib version

Matplotlib version: 2.2.2
Python version: 3.6

@anntzer

Bisects to 722f405, so quite a long time ago actually! ping @anntzer

Something like

diff --git a/lib/matplotlib/lines.py b/lib/matplotlib/lines.py
index cc8c61754..738a2396f 100644
--- a/lib/matplotlib/lines.py
+++ b/lib/matplotlib/lines.py
@@ -745,6 +745,8 @@ class Line2D(Artist):
             subslice = slice(max(i0 - 1, 0), i1 + 1)
             self.ind_offset = subslice.start
             self._transform_path(subslice)
+        else:
+            subslice = None
 
         transf_path = self._get_transformed_path()
 
@@ -804,7 +806,15 @@ class Line2D(Artist):
             gc.set_antialiased(self._antialiased)
 
             marker = self._marker
-            tpath, affine = transf_path.get_transformed_points_and_affine()
+
+            # Markers *must* be drawn ignoring the drawstyle.
+            with cbook._setattr_cm(
+                    self, _drawstyle="default", _transformed_path=None):
+                self.recache()
+                if subslice:
+                    self._transform_path(subslice)
+                tpath, affine = (self._get_transformed_path()
+                                 .get_transformed_points_and_affine())
             if len(tpath.vertices):
                 # subsample the markers if markevery is not None
                 markevery = self.get_markevery()

fixes the issue, but that's pretty horrible from a performance point of view. Don't have the time to turn this into a proper PR right now.

great, thanks!

Navigation Menu

Repository navigation

wrong x-position of marker with drawstyle='steps-xxx' #11031

Bug report

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

wrong x-position of marker with drawstyle='steps-xxx' #11031

Description

Bug report

Activity

dstansby commented on Apr 12, 2018

anntzer commented on Apr 12, 2018

pierre-haessig commented on Jun 11, 2018

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions