matplotlib-devel Mailing List for matplotlib

Brad Chivari wrote:
> SUBJECT:
> Filtering out 0 bar height/width not working
> 
> FILE:
> matplotlib/ trunk/ matplotlib/ lib/ matplotlib/ axes.py
> 
> PROBLEM:
> xmin = np.amin(width[width!=0]) # filter out the 0 width rects
> ymin = np.amin(height[height!=0]) # filter out the 0 height rects
> 
> These aren't using proper python list comprehension and don't work as expected (for me anyway).
> 
> SOLUTION:
> Shouldn't they be something like:
> xmin = np.amin([w for w in width if w != 0])
> ymin = np.amin([h for h in height if h != 0])
> 
> Once I changed them they seem to work properly.

I have applied a slight modification to your fix, which seems to me to 
be the right approach, given that the bar implementation is based on 
lists rather than ndarrays.

Thanks.

Eric

Ryanitus wrote:
> I may have found a bug in the __setitem__ method of the maxdict class.
> 

Fixed--thanks for pointing it out.

Eric

> Since a dictionary is a mapping class, if an item is set that already
> exists, it overwrites the previous.  However, you are still appending that
> item to _killkeys regardless.
> 
> In the case where 2 items with the same key were added and later removed due
> to size constraints, the line in bold would throw an exception on the second
> call.
> 
> By checking for the key existance, you handle that situation gracefully. 
> The item can still be removed from _killkeys, since it's the next to go
> anyway.
> 
> Ryan
> 
> Original code (circa line 776):
> 
>         if len(self)>=self.maxsize:
>             del self[self._killkeys[0]]
>             del self._killkeys[0]
>         dict.__setitem__(self, k, v)
>         self._killkeys.append(k)
> 
> Possible solution:
> 
>         if len(self)>=self.maxsize:
>             if self.has_key(self._killkeys[0]):
>                 del self[self._killkeys[0]]
>             del self._killkeys[0]
>         dict.__setitem__(self, k, v)
>         self._killkeys.append(k)
> 
> 
> 

On Fri, Sep 18, 2009 at 7:01 PM, Andrew Straw <str...@as...> wrote:

> Maybe the MPL binary was built with a numpy svn version that had API
> incompatibilities with numpy releases?

I built the python2.6 OSX 0.99.1rc1 binary with the numpy from the
1.3.0 dmg installer from the sf site, intentionally not using my local
numpy from svn since I understand there are some ABI incompatibilities
between 1.3.0 and 1.4.0svn and I wanted to target the latest release.
 What version are you using David?

JDH

David Warde-Farley wrote:
> On 18-Sep-09, at 6:42 PM, David Warde-Farley wrote:
>   
>>   File "/Library/Frameworks/Python.framework/Versions/2.6/lib/
>> python2.6/site-packages/ipython-0.10-py2.6.egg/IPython/Shell.py", line
>> 627, in __init__
>>     user_ns,user_global_ns,b2 =
>> self._matplotlib_config(name,user_ns,user_global_ns)
>>   File "/Library/Frameworks/Python.framework/Versions/2.6/lib/
>> python2.6/site-packages/ipython-0.10-py2.6.egg/IPython/Shell.py", line
>> 556, in _matplotlib_config
>>     import matplotlib.pylab as pylab
>>   File "/Library/Frameworks/Python.framework/Versions/2.6/lib/
>> python2.6/site-packages/matplotlib/pylab.py", line 206, in <module>
>>     from matplotlib import mpl  # pulls in most modules
>>   File "/Library/Frameworks/Python.framework/Versions/2.6/lib/
>> python2.6/site-packages/matplotlib/mpl.py", line 1, in <module>
>>     from matplotlib import artist
>>   File "/Library/Frameworks/Python.framework/Versions/2.6/lib/
>> python2.6/site-packages/matplotlib/artist.py", line 5, in <module>
>>     from transforms import Bbox, IdentityTransform, TransformedBbox,
>> TransformedPath
>>   File "/Library/Frameworks/Python.framework/Versions/2.6/lib/
>> python2.6/site-packages/matplotlib/transforms.py", line 34, in  
>> <module>
>>     from matplotlib._path import affine_transform
>> ImportError: numpy.core.multiarray failed to import
>>     
>
> Looking a little harder at this (after downloading the source) I'm  
> positively at a loss, since multiarray doesn't seem to be explicitly  
> imported anywhere in matplotlib. Furthermore, importing  
> numpy.core.multiarray from ipython works just fine.
>
> Moreover, I just built matplotlib-0.99.1rc1 from source and the  
> problem doesn't manifest. _Weird_. Any idea what's going on?
>   
Maybe the MPL binary was built with a numpy svn version that had API
incompatibilities with numpy releases?