Bugfix in ScalarFormatter._set_offset(); doc tweak in transforms.py

efiring · efiring · commit 5d523b44aa4e · 2006-05-14T07:50:54.000Z
svn path=/trunk/matplotlib/; revision=2389
diff --git a/lib/matplotlib/ticker.py b/lib/matplotlib/ticker.py
@@ -304,8 +304,9 @@ def _set_offset(self, range):
         # offset of 20,001 is 20,000, for example
         locs = self.locs
 
-        if locs is None or not len(locs):
+        if locs is None or not len(locs) or range == 0:
             self.offset = 0
+            return
         ave_loc = mean(locs)
         if ave_loc: # dont want to take log10(0)
             ave_oom = math.floor(math.log10(mean(absolute(locs))))
diff --git a/lib/matplotlib/transforms.py b/lib/matplotlib/transforms.py
@@ -197,8 +197,7 @@
     manipulations for which the PBox class was made.
 
     Bbox uses LazyValues grouped in pairs as 'll' and 'ur' Point objects;
-    PBox uses a 4-element list; it is subclassed from the python list, so
-    it could be used directly in place of a list as an Axes._position.
+    PBox uses a 4-element list, subclassed from the python list.
 
     Bbox and PBox methods are mostly quite different, reflecting their
     different original purposes.  Similarly, the CXX implementation of
