matplotlib-users Mailing List for matplotlib

Hi,
 In matplotlib, is there a way to check whether a window related to a
figure has been closed by user?
x=[1,2,3]
fig = plt.figure()
fig.show()
plt.plot(x,x)
plt.draw()
later in the code, I want to make draw some other line, possibly on the
same figure. I can switch to the preivious figure by
plt.figure(fig.number)
but it seems that even if fig has been closed by the user, this will still
work because the information stored in fig is still accessible. Then, this
fig.number would actually point some other picture.
So, is there any state information in fig that would tell whether the
window of fig has been closed by user?
Thanks
Qian

Hi everyone,
I'm trying to make a zoom of a single plot. The problem is that the zoomed
axes yticks overlap with the yaxes so I tried to use the bbox_to_anchor
option in the zoom:
axins = zoomed_inset_axes(ax,
3,bbox_to_anchor=(0.4,0.4),bbox_transform=ax.transAxes, loc=2)
with different (x,y) parameters, but it's not working.
I also tried to move the figure by hand:
box = axins.get_position()
print box
axins.set_position([box.x0+0.5, box.y0,box.width, box.height])
box = axins.get_position()
print box
and I see that box numbers changing, but not in the plot.
Can anyone guess what I am doing wrong?
Thank you all!
PS: The full code for this plot is:
def plotvsdata(data,z,dl,nu):
 """
 Plot of z vs. Dl for different nu values
 """
 fig = p.figure(figsize=(12,8))
 ax = p.subplot(111)
 # FLRW light distance
 ax.plot(z[0],Dl_lflrw(z[0],0.3,0.7),'k--',label=r'Dl $\Lambda$FLRW')
 # mean data values
 ax.plot(data.redshift,data.distance,'r.',label='SCP data')
 p.xlabel('z',fontsize=18)
 p.ylabel('Luminosity distance',fontsize=18)
 p.grid(True)
 #axins = zoomed_inset_axes(ax,3,loc=2,bbox_to_anchor=(0.5,0.5))
 #axins = zoomed_inset_axes(ax,3,loc=2,bbox_to_anchor=(0.4,0.4)
,bbox_transform=ax.transAxes)
 axins = zoomed_inset_axes(ax,
3,bbox_to_anchor=(1,1),bbox_transform=ax.transAxes, loc=2)
 #box = axins.get_position()
 #print box
 #axins.set_position([box.x0+0.5, box.y0,box.width, box.height])
 #box = axins.get_position()
 #print box
 # Kantowski equation
 for i in range(len(z)):
 ax.plot(z[i],dl[i],label=r'$\nu={0:1.2f}$'.format(nu[i]))
 axins.plot(z[i],dl[i])
 axins.plot(data.redshift,data.distance,'r.', label='SCP data')
 #box = ax.get_position()
 #ax.set_position([box.x0,box.y0,box.width*0.8,box.height])
 ax.legend(loc=2,bbox_to_anchor=(1,1))
 x,y = (0.86,1.3)
 x2,y2 = (1,1.7)
 axins.set_xlim(x,x2)
 axins.set_ylim(y,y2)
 axins.tick_params(axis='both',labelsize=12)
 axins.xaxis.set_major_locator(MaxNLocator(2))
 axins.yaxis.set_major_locator(MaxNLocator(3))
 #box = axins.get_position()
 #print box
 #axins.set_position([box.x0+0.5, box.y0,box.width, box.height])
 #box = axins.get_position()
 #print box
 # draw a bbox of the region to zoom and connecting lines between it
 # and the zoomed box.
 mark_inset(ax,axins,loc1=1,loc2=3, fc="none", ec="0.5")
 #p.xticks(visible=False)
 #p.yticks(visible=False)
 p.draw()
 p.show()

I tried a few things and found out that doing a pause works.
So why does a pause work, but a draw() or show() does not?
This all on Windows using the standard PythonXY installation.
Here is the code that works (testimage.py):
from pylab import *
c = ones((10,20))
ax = imshow(c)
pause(0.01)
print ax.get_axes().get_position()
Running from IPython:
run testimage
Bbox(array([[ 0.125 , 0.24166667],
 [ 0.9 , 0.75833333]]))
On Wed, Aug 1, 2012 at 8:56 PM, Stan West <sta...@nr...> wrote:
> **
>
> *From:* Mark Bakker [mailto:ma...@gm...]
> *Sent:* Monday, July 30, 2012 05:54
>
> Hello List,
>
> I am trying to determine the position of the axes after an imshow and am
> having problems.
> I get a different answer on my Mac (the correct answer) than Windows (the
> wrong answer).
>
> [...]
>
> Any thoughts? When I type the commands in at the IPython prompt it works
> most of the time (on Windows), but it never works when running the file.
> What in the world could be different?
>
> mp version 1.1.0 on both systems.
>
> Are you using the same backend on both systems?
>
> Perhaps when you run the script within IPython on your Windows system, the
> show() call is not triggering a draw. You can force drawing by calling
> draw() before show(). I hope that helps.
>

On 7 August 2012 03:10, Jeff Whitaker <js...@fa...> wrote:
> Available for download at
> https://sourceforge.net/projects/matplotlib/files/matplotlib-toolkits/basemap-1.0.5/
Just a note to anyone experiencing problems building against GEOS
3.3.3 installed from the ubuntu-gis-unstable PPA on Ubuntu. The naming
convention for the GEOS shared library has changed and I had to apply
the patch below as a workaround:
diff --git a/setup.py b/setup.py
index 216ddf6..dc6d916 100644
--- a/setup.py
+++ b/setup.py
@@ -82,7 +82,7 @@ else:
 library_dirs=geos_library_dirs,
 runtime_library_dirs=geos_library_dirs,
 include_dirs=geos_include_dirs,
- libraries=['geos_c','geos']))
+ libraries=['geos_c','geos-3.3.3']))
 # Specify all the required mpl data
 # create pyproj binary datum shift grid files.
Apparently this is an acceptable practice on Debian systems
(http://www.debian.org/doc/debian-policy/ch-sharedlibs.html#s-shlibs),
but no time to dream up a more robust fix right now..
Cheers,
Scott

Available for download at
https://sourceforge.net/projects/matplotlib/files/matplotlib-toolkits/basemap-1.0.5/
Thanks to Christoph Gohlke for making windows installers.
Highlights:
 - new 'latlon' keyword for contour,contourf,pcolor,pcolormesh,quiver,barb methods that automatically converts x, y from longitude/latitude to map projection coordinates, and for global projections shifts the data longitudinally to fit in map projection region (see shiftdata.py example).
 - fixed a serious bug that caused a bogus mask to be applied to the data when the contour method was called (did not affect contourf).
full Changelog:
version 1.0.5 (git tag v1.0.5rel)
--------------------------------
* fix bug triggered when drawlsmask method called more than once.
* fix error in contour method that caused a bogus mask to be applied
 to the data (issue 58).
* fix further corner cases with splitting of parallels that cross
 the dateline (issue 40).
* add latlon keyword to plotting methods. If latlon=True, x and y
 values are assumed to longitudes and latitudes in degrees. The
 data and longitudes are shifted to the map projection region (for
 cylindrical and pseudo-cylindrical projections) using the shiftdata
 method, and lons/lats are converted to map projection coords. Default
 value is False. Addresses issue 54. New example shiftdata.py added
 to illustrate usage.
* fix bluemarble and warpimage methods to account for change in orientation
 of arrays returned to matplotlib's pil_to_array (issue 51)
* fix glitch with drawing meridians and filling coastline polygons with
 omerc projection that includes pole.
Regards,
Jeff