matplotlib-users Mailing List for matplotlib

Adam,

I agree that the Circle ended up with a rectangle in the legend, which I wouldn't think of as the expected response.  Would the following work for your purposes?

figure()
p,=plot(0,0,marker='o',ls='',color='red')
legend([p], ["Red Rectangle"],numpoints=1)
p.remove()
draw()

-Sterling

On Apr 11, 2014, at 8:14AM, Adam Hughes wrote:

> Hi Paul, 
> 
> I tried out the legend proxy artist, and it works for rectangles in the legend, but I can't seem to get a Circle to appear in the legend, which I presume should be:
> 
> p = Circle((0, 0), fc="r")
> legend([p], ["Red Rectangle"])
> 
> 
> On Wed, Apr 9, 2014 at 2:20 PM, Adam Hughes <hug...@gm...> wrote:
> Thanks Paul, I will try it out.
> 
> 
> On Wed, Apr 9, 2014 at 12:21 PM, Paul Hobson <pmh...@gm...> wrote:
> 
> 
> 
> On Wed, Apr 9, 2014 at 9:00 AM, Adam Hughes <hug...@gm...> wrote:
> Thanks.  That's probably the way I'll go.  At first, I thought creating separate legend markers and removing them from the plot seemed hacky, but I guess there's no way that matplotlib could know which legend size I want.  I wonder if there'd be any interest in a PR to add a keyword to legend to handle this situation?
> 
> Why not just work the other way around with proxy artists. IOW, make the artists but never add them to the plot.
> 
> http://matplotlib.org/users/legend_guide.html?highlight=proxy%20artists#using-proxy-artist
> (works with Line2D artists)
> 
> -p
> 
>  
> 
> On Wed, Apr 9, 2014 at 1:44 AM, Sterling Smith <sm...@fu...> wrote:
> Adam,
> 
> I haven't investigated, but does the discussion of the legend marker at [1] help?
> 
> -Sterling
> 
> [1] https://www.mail-archive.com/mat...@li.../msg25200.html
> 
> On Apr 8, 2014, at 3:44PM, Adam Hughes wrote:
> 
> > Hello,
> >
> > I've been searching but can't seem to find this topic addressed (perhaps wrong search terms)
> >
> > Simply put, I have a scatter plot with variable size markers, and I'd like to have the markers all be a single size in the legend.  Is there a standard way to do this?
> >
> > Thanks.
> > ------------------------------------------------------------------------------
> > Put Bad Developers to Shame
> > Dominate Development with Jenkins Continuous Integration
> > Continuously Automate Build, Test & Deployment
> > Start a new project now. Try Jenkins in the cloud.
> > http://p.sf.net/sfu/13600_Cloudbees_______________________________________________
> > Matplotlib-users mailing list
> > Mat...@li...
> > https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
> 
> 
> ------------------------------------------------------------------------------
> Put Bad Developers to Shame
> Dominate Development with Jenkins Continuous Integration
> Continuously Automate Build, Test & Deployment
> Start a new project now. Try Jenkins in the cloud.
> http://p.sf.net/sfu/13600_Cloudbees
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
> 
> 
> 

I can confirm the inconsistency between behavior and documentation for 1.3.1.  The errorbar line gets the color of the line, not the marker.  Probably you should file a bug report on github.

-Sterling

On Apr 11, 2014, at 7:50AM, Oliver wrote:

> I apologize if this has been fixed already, I can only check different versions at home. However, the documentation of mpl 1.3.1. has the same information. So unless the code changed to reflect the documentation, this is still present.
> 
> When using errorbar, the documentation says the color of the errorbar lines will match with the color of the markers if ecolor=None. That’s not what I found. Apparently it takes over the color of the Line2D instance which interconnects the markers.
> 
> Short, Self Contained, Correct Example:
> 
> from pylab import *
> plt.ion()  # saves typing show
> 
> x = np.arange(10)
> y = np.random.rand(10)
> xerr, yerr = y/4., y/4.
> 
> # Markers in red, but errorlines assume the color of the "trendline" (default rcparams: blue).
> errorbar(x, y, yerr=yerr, mfc='r', marker='o', ecolor=None) 
> 
> # Errorlines get color green now - documentation not in line with results
> 
> figure(); errorbar(x, y, yerr=xerr, mfc='r', marker='o', ecolor=None, color='g')
> # Errorlines get color blue now, because it can be specified - expected behaviour
> figure(); errorbar(x, y, yerr=xerr, mfc='r', marker='o', ecolor='b', color='g')
> 
> Is this an oversight mistake?
> 
> ------------------------------------------------------------------------------
> Put Bad Developers to Shame
> Dominate Development with Jenkins Continuous Integration
> Continuously Automate Build, Test & Deployment 
> Start a new project now. Try Jenkins in the cloud.
> http://p.sf.net/sfu/13600_Cloudbees_______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Under Win7pro, I have tried to install matplotlib by running either 
matplotlib-1.3.0.win32-py3.3.exe or
matplotlib-1.3.1.win32-py3.3.exe. In each case, the installer tells 
pythion-3.3 is not in the registry (although
Anaconda was successfully installed previousle). A pop-up window then 
opens, asking me for the python
and installation directories. It proves impossible to write anything in 
this window.
Does anybody know what's wrong in my set up ?
TIA for any help
JP Grivet*

Hi Paul,

I tried out the legend proxy artist, and it works for rectangles in the
legend, but I can't seem to get a Circle to appear in the legend, which I
presume should be:

p = Circle((0, 0), fc="r")
legend([p], ["Red Rectangle"])


On Wed, Apr 9, 2014 at 2:20 PM, Adam Hughes <hug...@gm...> wrote:

> Thanks Paul, I will try it out.
>
>
> On Wed, Apr 9, 2014 at 12:21 PM, Paul Hobson <pmh...@gm...> wrote:
>
>>
>>
>>
>> On Wed, Apr 9, 2014 at 9:00 AM, Adam Hughes <hug...@gm...>wrote:
>>
>>> Thanks.  That's probably the way I'll go.  At first, I thought creating
>>> separate legend markers and removing them from the plot seemed hacky, but I
>>> guess there's no way that matplotlib could know which legend size I want.
>>>  I wonder if there'd be any interest in a PR to add a keyword to legend to
>>> handle this situation?
>>>
>>
>> Why not just work the other way around with proxy artists. IOW, make the
>> artists but never add them to the plot.
>>
>>
>> http://matplotlib.org/users/legend_guide.html?highlight=proxy%20artists#using-proxy-artist
>> (works with Line2D artists)
>>
>> -p
>>
>>
>>
>>>
>>> On Wed, Apr 9, 2014 at 1:44 AM, Sterling Smith <sm...@fu...>wrote:
>>>
>>>> Adam,
>>>>
>>>> I haven't investigated, but does the discussion of the legend marker at
>>>> [1] help?
>>>>
>>>> -Sterling
>>>>
>>>> [1]
>>>> https://www.mail-archive.com/mat...@li.../msg25200.html
>>>>
>>>> On Apr 8, 2014, at 3:44PM, Adam Hughes wrote:
>>>>
>>>> > Hello,
>>>> >
>>>> > I've been searching but can't seem to find this topic addressed
>>>> (perhaps wrong search terms)
>>>> >
>>>> > Simply put, I have a scatter plot with variable size markers, and I'd
>>>> like to have the markers all be a single size in the legend.  Is there a
>>>> standard way to do this?
>>>> >
>>>> > Thanks.
>>>> >
>>>> ------------------------------------------------------------------------------
>>>> > Put Bad Developers to Shame
>>>> > Dominate Development with Jenkins Continuous Integration
>>>> > Continuously Automate Build, Test & Deployment
>>>> > Start a new project now. Try Jenkins in the cloud.
>>>> >
>>>> http://p.sf.net/sfu/13600_Cloudbees_______________________________________________
>>>> > Matplotlib-users mailing list
>>>> > Mat...@li...
>>>> > https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>>>
>>>>
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Put Bad Developers to Shame
>>> Dominate Development with Jenkins Continuous Integration
>>> Continuously Automate Build, Test & Deployment
>>> Start a new project now. Try Jenkins in the cloud.
>>> http://p.sf.net/sfu/13600_Cloudbees
>>> _______________________________________________
>>> Matplotlib-users mailing list
>>> Mat...@li...
>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>>
>>>
>>
>

I apologize if this has been fixed already, I can only check different
versions at home. However, the documentation of mpl
1.3.1<http://matplotlib.org/1.3.1/api/pyplot_api.html#matplotlib.pyplot.errorbar>.
has the same information. So unless the code changed to reflect the
documentation, this is still present.

When using errorbar, the documentation says the color of the errorbar lines
will match with the color of the markers if ecolor=None. That’s not what I
found. Apparently it takes over the color of the Line2D instance which
interconnects the markers.

Short, Self Contained, Correct Example:

from pylab import *
plt.ion()  # saves typing show

x = np.arange(10)
y = np.random.rand(10)
xerr, yerr = y/4., y/4.

# Markers in red, but errorlines assume the color of the "trendline"
(default rcparams: blue).
errorbar(x, y, yerr=yerr, mfc='r', marker='o', ecolor=None)

# Errorlines get color green now - documentation not in line with results

figure(); errorbar(x, y, yerr=xerr, mfc='r', marker='o', ecolor=None, color='g')
# Errorlines get color blue now, because it can be specified -
expected behaviour
figure(); errorbar(x, y, yerr=xerr, mfc='r', marker='o', ecolor='b', color='g')

Is this an oversight mistake?