Rely on normal autoscaling for stepfilled hist.

efiring · efiring · commit fbf97d4be4c2 · 2016-11-24T11:24:20.000-10:00
diff --git a/lib/matplotlib/axes/_axes.py b/lib/matplotlib/axes/_axes.py
@@ -6334,7 +6334,7 @@ def _normalize_input(inp, ename='input'):
 
                 y[0], y[-1] = minimum, minimum
             else:
-                minimum = np.min(bins)
+                minimum = 0  # np.min(bins)
 
             if align == 'left' or align == 'center':
                 x -= 0.5*(bins[1]-bins[0])
@@ -6386,14 +6386,6 @@ def _normalize_input(inp, ename='input'):
             # we return patches, so put it back in the expected order
             patches.reverse()
 
-            # adopted from adjust_x/ylim part of the bar method
-            if orientation == 'horizontal':
-                xmax = self.dataLim.intervalx[1]
-                self.dataLim.intervalx = (0, xmax)
-            elif orientation == 'vertical':
-                ymax = self.dataLim.intervaly[1]
-                self.dataLim.intervaly = (0, ymax)
-
         self.set_autoscalex_on(_saved_autoscalex)
         self.set_autoscaley_on(_saved_autoscaley)
         self.autoscale_view()
