matplotlib-users Mailing List for matplotlib

Charlie Moad wrote:
>    I guess I find it strange that I always get color output when
> compiling matplotlib now.  I know it is using numpy's distutils
> because I have done some printing in the setup.py before.  I wonder if
> having numpy and family installed with setuptools is doing this.  Any
> clues?

No, you're right, numpy.distutils is getting pulled in through the backdoor. I'm
entering this as a bug in numpy.

-- 
Robert Kern
rob...@gm...

"In the fields of hell where the grass grows high
 Are the graves of dreams allowed to die."
  -- Richard Harter

   I guess I find it strange that I always get color output when
compiling matplotlib now.  I know it is using numpy's distutils
because I have done some printing in the setup.py before.  I wonder if
having numpy and family installed with setuptools is doing this.  Any
clues?

On 2/28/06, Robert Kern <rob...@gm...> wrote:
> Charlie Moad wrote:
> > Numpy and Scipy highjack distutils now, and this may be something they
> > add in.  Do you ave numpy installed?  That looks like a valid option
> > though.
>
> Neither numpy nor scipy do anything to the main distutils. numpy.distutil=
s
> (replacing the old scipy_distutils) extends distutils only if you import =
it;
> nothing happens to distutils if you just import numpy. matplotlib doesn't=
 import
> numpy.distutils at all.
>
> --
> Robert Kern
> rob...@gm...
>
> "In the fields of hell where the grass grows high
>  Are the graves of dreams allowed to die."
>   -- Richard Harter
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by xPML, a groundbreaking scripting langua=
ge
> that extends applications into web and mobile media. Attend the live webc=
ast
> and join the prime developer group breaking into this new coding territor=
y!
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D110944&bid=3D241720&dat=
=3D121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Charlie Moad wrote:
> Numpy and Scipy highjack distutils now, and this may be something they
> add in.  Do you ave numpy installed?  That looks like a valid option
> though.

Neither numpy nor scipy do anything to the main distutils. numpy.distutils
(replacing the old scipy_distutils) extends distutils only if you import it;
nothing happens to distutils if you just import numpy. matplotlib doesn't import
numpy.distutils at all.

-- 
Robert Kern
rob...@gm...

"In the fields of hell where the grass grows high
 Are the graves of dreams allowed to die."
  -- Richard Harter

On Sun, 26 Feb 2006, Charlie Moad wrote:

> Numpy and Scipy highjack distutils now, and this may be something they
> add in.  Do you ave numpy installed?  That looks like a valid option
> though.

erm. I'm not entirely sure. I've somehow broken my "yast" in suse whilst 
trying to install matplotlib and now I cant see what's installed.
[ any help on how to fix a malloc memory corruption in yast greatfully 
accepted ;-) ]
I ran python and tried to import numpy and scipy but that didnt work, so 
I'm assuming I dont have them. (is there a better way to tell? - I'm sorry 
if I sound very ignorant, but I've only been learning python a few weeks).

thanks alot.
Sue

> > I'm having problems compiling matplotlib. I'm running suse 9.3 (linux) and
> > have python:
> > Python 2.4.1 (#1, Sep 13 2005, 00:39:20)
> >
> > I downloaded matplotlib-0.87.tar.gz, and have not changed the setup.py or
> > setupext.py at all. I think I have all the graphics things installed.
> >
> > The error message goes:
> > >>> [snip]
> > building 'matplotlib._agg' extension
> > creating build/temp.linux-i686-2.4
> > creating build/temp.linux-i686-2.4/src
> > creating build/temp.linux-i686-2.4/agg23
> > creating build/temp.linux-i686-2.4/agg23/src
> > gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -march=i586 -mtune=i686
> > -fmessage-length=0 -Wall -D_FORTIFY_SOURCE=2 -g -I/usr/X11R6/include -fPIC
> > -Iagg23/include -Isrc -Iswig -I/usr/include/python2.4 -c src/agg.cxx -o
> > build/temp.linux-i686-2.4/src/agg.o
> > cc1plus: error: invalid option `tune=i686'
> > error: command 'gcc' failed with exit status 1
> >
> > >>  gcc --version
> > gcc (GCC) 3.3.5 20050117 (prerelease) (SUSE Linux)
> > Copyright (C) 2003 Free Software Foundation, Inc.
> >
> > I cant find anywhere in the matplotlib-0.87 directory structure that sets
> > this "tune=i686". I have searched all the files in the directory structure
> > for "tune" and "i686", so I'm guessing it's set somewhere else, but I dont
> > know where.
> > Can anyone help?
> >
> > thanks
> > Sue
> >
> >
> >
> > -------------------------------------------------------
> > This SF.Net email is sponsored by xPML, a groundbreaking scripting language
> > that extends applications into web and mobile media. Attend the live webcast
> > and join the prime developer group breaking into this new coding territory!
> > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
> > _______________________________________________
> > Matplotlib-users mailing list
> > Mat...@li...
> > https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> >
> 
> 

|------------------------------------------------------------------|
| email    su...@po...                           |
| WWW      http://www.pooh-room.freeserve.co.uk                    |
|------------------------------------------------------------------|

On Tue, 28 Feb 2006, Steve Schmerler apparently wrote:=20
> In [39]: a =3D arange(0, 0.005, 0.00001); len(a) Out[39]:=20
> 500=20
> In [40]: a =3D arange(0, 0.005, 0.000001); len(a)=20
> Out[40]: 5001=20
> Shouldn't  len(a)%10 =3D=3D 0 in these cases?=20


Use numpy.linspace:

linspace(start, stop, num=3D50, endpoint=3DTrue, retstep=3DFalse)
    Return evenly spaced numbers.

    Return 'num' evenly spaced samples from 'start' to 'stop'.  If
    'endpoint' is True, the last sample is 'stop'. If 'retstep' is
    True then return the step value used.

Cheers,
Alan Isaac

PS http://docs.python.org/tut/node16.html

Steve Schmerler wrote:
> Hi
> 
> I've a machine where I used Numeric until now. Does the following
> behaviour also occur with numpy?
> 
> 
> In [38]: Numeric.__version__
> Out[38]: '23.8'
> 
> In [39]: a = arange(0, 0.005, 0.00001); len(a)
> Out[39]: 500
> 
> In [40]: a = arange(0, 0.005, 0.000001); len(a)
> Out[40]: 5001
> 
> Shouldn't  len(a)%10 == 0 in these cases?

Yes. Floating point is weird.

http://projects.scipy.org/scipy/numpy/ticket/8

Use numpy.linspace() for reliable results. I think matplotlib exposes
linspace(), too.

-- 
Robert Kern
rob...@gm...

"In the fields of hell where the grass grows high
 Are the graves of dreams allowed to die."
  -- Richard Harter

Hi

I've a machine where I used Numeric until now. Does the following 
behaviour also occur with numpy?


In [38]: Numeric.__version__
Out[38]: '23.8'

In [39]: a = arange(0, 0.005, 0.00001); len(a)
Out[39]: 500

In [40]: a = arange(0, 0.005, 0.000001); len(a)
Out[40]: 5001

Shouldn't  len(a)%10 == 0 in these cases?

cheers,
steve

-- 
Random number generation is the art of producing pure gibberish as 
quickly as possible.

I don't know if it something wanted but it's very useful for me. For 
example sometime to improve the contrast between two values, I plot the 
log of the image but I don't want the value of the colorbar in log but 
linear or another sample (problably not good but I didn't find any other 
way to do it), I want plot the data from yellow to red, I'm using the 
hot colormat to do it but I didn't find any way to do the opposite so I 
began by plotting the data with the normal value, I plot the colorbar 
and I plot another time the opposite of the data, after with gimp I turn 
the colorbar and my scale is ok.

It's bad and dirty but it's working. I didn't find any otther way to do 
it properly.

N.

Jon Fox wrote:
> I've been working on a matplotlib embedded gtk app that updates an image
> (made with imshow) with a colorbar and I've noticed that although I can
> update the image with an image.set_data(new_array) that the colormapping
> and colorbar don't change. Is this the intended behaviour?  I guess it
> makes perfect sense if the image doesn't change much. 
>
> I am using a figure.clear() and a completely new axes.imshow() each time
> I redraw the plot. 
>
> -- Jon
>
>
>   

I've been working on a matplotlib embedded gtk app that updates an image
(made with imshow) with a colorbar and I've noticed that although I can
update the image with an image.set_data(new_array) that the colormapping
and colorbar don't change. Is this the intended behaviour?  I guess it
makes perfect sense if the image doesn't change much. 

I am using a figure.clear() and a completely new axes.imshow() each time
I redraw the plot. 

-- Jon


-- 
.*. Dr. Jon R. Fox
..* http://www.drfox.com
*** jo...@dr...

>>>>> "Christian" == Christian Meesters <mee...@un...> writes:

    Christian> Hi, Is there a way to get a legend without a border?
    Christian> (Using pylab.legend().)  (Sorry if this is all too
    Christian> easy, but I couldn't find this anywhere.)

Then it must not be *too* easy to find :-)

The legend command returns a legend instance

  leg = ax.legend(something...)

and you can find the legend methods at

  http://matplotlib.sourceforge.net/matplotlib.legend.html

Specifically the "draw_frame" method

  leg.draw_frame(False)

Should help,
JDH

>>>>> "Jeff" == Jeff Peery <jef...@ya...> writes:

    Jeff> Hello, I am having trouble with scatter plots. I created a
    Jeff> scatter plot of just one point: s1 = scatter([1,],[1,])
  
    Jeff>   now I want to move this point after its creation and I'm
    Jeff> not sure how to do this. I'm been looking around in teh
    Jeff> documentation without any success. Is there something that
    Jeff> might be close to the following:

It is possible but not advisable.  scatter will be slower than plot
for this purpose, and since you have only one point there is not
reason to use a PolygonCollection, which is what scatter uses under
the hood.  You can just

  line, = ax.plot([1], [1])

and then call
  
    Jeff>   
  line.set_xydata([2,],[2,])

But if I were just doing one point, I would create my own
matplotlib.patches.Polygon and use ax.add_patch to add it to the axes
and then manipulate the polygon properties directly.

  mpl/examples> grep Polygon *
  integral_demo.py:from matplotlib.patches import Polygon
  integral_demo.py:poly = Polygon(verts, facecolor=0.8, edgecolor='k')
  poly_editor.py:from matplotlib.patches import Polygon
  poly_editor.py:class PolygonInteractor:
  poly_editor.py:p = PolygonInteractor( circ)
  
  mpl/examples> grep Circle *
  picker_demo.py:    p = Circle(center, radius=.1)
  poly_editor.py:circ = Circle((.5,.5),.5)


JDH

Hi,

Is there a way to get a legend without a border? (Using pylab.legend().)
(Sorry if this is all too easy, but I couldn't find this anywhere.)

TIA
Christian

Hello, I am having trouble with scatter plots. I created a scatter plot of just one point:
  
  s1    = scatter([1,],[1,])
  
  now I want to move this point after its creation and I'm not sure how  to do this. I'm been looking around in teh documentation without any  success. Is there something that might be close to the following:
  
  s1.set_xydata([2,],[2,])
  
  I can't just clear my axes and redraw the scatter point because it  takes too long (there is lots of other stuff on the figure). so I just  want to move teh position of the scatter and then do a draw().
  
  any help would be much appreciated! thanks!
  
  Jeff

mat...@li... wrote:  Send Matplotlib-users mailing list submissions to
 mat...@li...

To subscribe or unsubscribe via the World Wide Web, visit
 https://lists.sourceforge.net/lists/listinfo/matplotlib-users
or, via email, send a message with subject or body 'help' to
 mat...@li...

You can reach the person managing the list at
 mat...@li...

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Matplotlib-users digest..."


Today's Topics:

   1. Re: visualization tool (Christopher Barker)
   2. Re: Bug in WXAgg on OS X and Matplotlib 0.87 (Samuel M. Smith)
   3. Re: Packaging matplotlib for Fedora Extras (Orion Poplawski)
   4. switch to subversion tomorrow Tuesday 28 Feb (Andrew Straw)
   5. Re: Packaging matplotlib for Fedora Extras (Christopher Barker)
   6. Re: switch to subversion tomorrow Tuesday 28 Feb (Vincent Favre-Nicolin)
   7. Re: Packaging matplotlib for Fedora Extras (Orion Poplawski)

--__--__--

Message: 1
Date: Mon, 27 Feb 2006 09:28:30 -0800
From: Christopher Barker 
Subject: Re: [Matplotlib-users] visualization tool
To: matplotlib-users 
Organization: NOAA/HAZMAT

> On Sat, 25 Feb 2006, Ryan Krauss wrote:

>> Can anyone recommend a python package for doing this kind of animation?

If you want to use wxPython, then wx.lib.floatcanvas would make this 
pretty easy.

-Chris



-- 
Christopher Barker, Ph.D.
Oceanographer
                                       
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...


--__--__--

Message: 2
Cc: "Samuel M. Smith" ,
 mat...@li...
From: "Samuel M. Smith" 
Subject: Re: [Matplotlib-users] Bug in WXAgg on OS X and Matplotlib 0.87
Date: Mon, 27 Feb 2006 10:34:02 -0700
To: Christopher Barker 

Onsi Fakhouri had the same problem and found a workaround. But I  
appreciate your help
the WX_AGG environment variable fixed one problem and your  
information on the linking tool should prove useful
to verify that things are getting build correctly.

See my post today titled "Workaround for 0.87 WXAgg bug".
Apparently according to Onsi
  there is a bug in the C code, memory is not being allocated correctly.


On 27 Feb, 2006, at 10:24, Christopher Barker wrote:

> Sam,
>
> I wish I could be more help,. but maybe I can help steer your  
> efforts  bit:
>
>> /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/ 
>> site-packages/matplotlib/backends/backend_wxagg.py in draw(self,  
>> repaint)
>>      60         FigureCanvasAgg.draw(self)
>>      61
>> ---> 62         self.bitmap = _convert_agg_to_wx_bitmap 
>> (self.get_renderer(), None)
>>      63         if repaint:
>>      64             self.gui_repaint()
>> MemoryError: _wxagg.convert_agg_to_wx_bitmap(): could not create  
>> the wx.Bitmap
>
> This sure looks like your error occurs when MPL is trying to  
> convert from the wxagg internal image storage to a wxBitmap. As a  
> test, I'd make sure the pure Agg back-end works fine (have it  
> create a PNG). If so then font,s etc have nothing to do withyour  
> trouble...it's all about the agg -> wx conversion. that's why I  
> thought it could be caused by liniking to the wrong version of wx.
>
> to make sure you have linked to the right version, you can use:
>
> otool -L
>
> on the wxagg libs that MPL builds. That should tell you which wx  
> libs you've actuallylinked to.
>
>
> With any luck, I'll have a chance to try to get this working on my  
> system this week, and then maybe I'll be more help.
>
> -Chris
>
>
>
>
>
>
> -- 
> Christopher Barker, Ph.D.
> Oceanographer
>                                       
> NOAA/OR&R/HAZMAT         (206) 526-6959   voice
> 7600 Sand Point Way NE   (206) 526-6329   fax
> Seattle, WA  98115       (206) 526-6317   main reception
>
> Chr...@no...

**********************************************************************
Samuel M. Smith Ph.D.
2966 Fort Hill Road
Eagle Mountain, Utah 84043
801-768-2768 voice
801-768-2769 fax
**********************************************************************
"The greatest source of failure and unhappiness in the world is
giving up what we want most for what we want at the moment"
**********************************************************************



--__--__--

Message: 3
Date: Mon, 27 Feb 2006 10:38:59 -0700
From: Orion Poplawski 
To: John Hunter 
CC: mat...@li...,
        Travis Oliphant 
Subject: Re: [Matplotlib-users] Packaging matplotlib for Fedora Extras

John Hunter wrote:
> 
> w/o addressing your original question (perhaps someone else has an
> opinion there) I suggest you wait until the 0.87.1 bug fix release
> because there is a known problem with 0.87 (images are broken).

When will this be?

-- 
Orion Poplawski
System Administrator                   303-415-9701 x222
Colorado Research Associates/NWRA      FAX: 303-415-9702
3380 Mitchell Lane, Boulder CO 80301   http://www.co-ra.com


--__--__--

Message: 4
Date: Mon, 27 Feb 2006 12:30:43 -0800
From: Andrew Straw 
To: mat...@li...,
 matplotlib development list 
Subject: [Matplotlib-users] switch to subversion tomorrow Tuesday 28 Feb

Hi all,

I will endeavor to switch matplotlib to a subversion repository (newly
available courtesy of SourceForge) tomorrow afternoon/evening
(US/Pacific time) unless I hear pleas from developers with unmerged changes.

I believe the new command for anonymous checkout of matplotlib will be:
svn co https://svn.sourceforge.net/svnroot/matplotlib/trunk/matplotlib

I'll post the developer checkout command on the matplotlib-devel list
when I figure it out.

Cheers!
Andrew


--__--__--

Message: 5
Date: Mon, 27 Feb 2006 13:52:26 -0800
From: Christopher Barker 
Subject: Re: [Matplotlib-users] Packaging matplotlib for Fedora Extras
To: or...@co...
Cc: mat...@li...
Organization: NOAA/HAZMAT

or...@co... wrote:
> I'm the maintainer for python-matplotlib in Fedora Extras.

Thanks for doing that, by the way.

 >  Is numpy fully functional

I think not quite, and it does change a bit with each release.

> the way of the future

Yes, it is the way of the future, but the future is not quite now.

> I have python-numeric as the required engine at runtime.

Just make it clear:

1) The binary is build so that it will work with all three

2) The question is what is defined as a dependency when you install it, 
and what is set as default in matplotlibrc.

If so, I'd stick with Numeric as the default. Those of us on the 
Bleeding Edge of Numpy can still use it, but people's older code should 
"just work"

As a rule of thumb, I'd wait until Travis calls numpy version 1.0 before 
changing the default.

-Chris


-- 
Christopher Barker, Ph.D.
Oceanographer
                                       
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...


--__--__--

Message: 6
From: Vincent Favre-Nicolin 
To: mat...@li...
Subject: Re: [Matplotlib-users] switch to subversion tomorrow Tuesday 28 Feb
Date: Mon, 27 Feb 2006 22:53:05 +0100

 Hi,

> I will endeavor to switch matplotlib to a subversion repository (newly
> available courtesy of SourceForge) tomorrow afternoon/evening
> (US/Pacific time) unless I hear pleas from developers with unmerged
> changes.
>
> I believe the new command for anonymous checkout of matplotlib will be:
> svn co https://svn.sourceforge.net/svnroot/matplotlib/trunk/matplotlib
>
> I'll post the developer checkout command on the matplotlib-devel list
> when I figure it out.

   It's actually the same command. Only, when committing changes you have t=
o=20
supply username and password using, e.g. --username and --password options.=
=20
But a subversion GUI (I'm happy with esvn for linux & windows) will do that=
=20
for you.
   What's really nice, there are no more delay between developers and=20
anonymous access.

 Vincent
=2D-=20
Vincent Favre-Nicolin
Universit=E9 Joseph Fourier
http://v.favrenicolin.free.fr
ObjCryst & Fox : http://objcryst.sourceforge.net


--__--__--

Message: 7
Date: Mon, 27 Feb 2006 14:57:05 -0700
From: Orion Poplawski 
To: Christopher Barker 
CC: mat...@li...
Subject: Re: [Matplotlib-users] Packaging matplotlib for Fedora Extras

Christopher Barker wrote:
> 
> Just make it clear:
> 
> 1) The binary is build so that it will work with all three
> 

Yup.  Or at least, that is the goal.

> 2) The question is what is defined as a dependency when you install it, 
> and what is set as default in matplotlibrc.
> 
> If so, I'd stick with Numeric as the default. Those of us on the 
> Bleeding Edge of Numpy can still use it, but people's older code should 
> "just work"

Will do.  Thanks!

-- 
Orion Poplawski
System Administrator                   303-415-9701 x222
Colorado Research Associates/NWRA      FAX: 303-415-9702
3380 Mitchell Lane, Boulder CO 80301   http://www.co-ra.com



--__--__--

_______________________________________________
Matplotlib-users mailing list
Mat...@li...
https://lists.sourceforge.net/lists/listinfo/matplotlib-users


End of Matplotlib-users Digest



		
---------------------------------
Yahoo! Mail
Bring photos to life! New PhotoMail  makes sharing a breeze. 

Christopher Barker wrote:
> 
> Just make it clear:
> 
> 1) The binary is build so that it will work with all three
> 

Yup.  Or at least, that is the goal.

> 2) The question is what is defined as a dependency when you install it, 
> and what is set as default in matplotlibrc.
> 
> If so, I'd stick with Numeric as the default. Those of us on the 
> Bleeding Edge of Numpy can still use it, but people's older code should 
> "just work"

Will do.  Thanks!

-- 
Orion Poplawski
System Administrator                   303-415-9701 x222
Colorado Research Associates/NWRA      FAX: 303-415-9702
3380 Mitchell Lane, Boulder CO 80301   http://www.co-ra.com

	Hi,

> I will endeavor to switch matplotlib to a subversion repository (newly
> available courtesy of SourceForge) tomorrow afternoon/evening
> (US/Pacific time) unless I hear pleas from developers with unmerged
> changes.
>
> I believe the new command for anonymous checkout of matplotlib will be:
> svn co https://svn.sourceforge.net/svnroot/matplotlib/trunk/matplotlib
>
> I'll post the developer checkout command on the matplotlib-devel list
> when I figure it out.

   It's actually the same command. Only, when committing changes you have t=
o=20
supply username and password using, e.g. --username and --password options.=
=20
But a subversion GUI (I'm happy with esvn for linux & windows) will do that=
=20
for you.
   What's really nice, there are no more delay between developers and=20
anonymous access.

	Vincent
=2D-=20
Vincent Favre-Nicolin
Universit=E9 Joseph Fourier
http://v.favrenicolin.free.fr
ObjCryst & Fox : http://objcryst.sourceforge.net

or...@co... wrote:
> I'm the maintainer for python-matplotlib in Fedora Extras.

Thanks for doing that, by the way.

 >  Is numpy fully functional

I think not quite, and it does change a bit with each release.

> the way of the future

Yes, it is the way of the future, but the future is not quite now.

> I have python-numeric as the required engine at runtime.

Just make it clear:

1) The binary is build so that it will work with all three

2) The question is what is defined as a dependency when you install it, 
and what is set as default in matplotlibrc.

If so, I'd stick with Numeric as the default. Those of us on the 
Bleeding Edge of Numpy can still use it, but people's older code should 
"just work"

As a rule of thumb, I'd wait until Travis calls numpy version 1.0 before 
changing the default.

-Chris


-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...

Hi all,

I will endeavor to switch matplotlib to a subversion repository (newly
available courtesy of SourceForge) tomorrow afternoon/evening
(US/Pacific time) unless I hear pleas from developers with unmerged changes.

I believe the new command for anonymous checkout of matplotlib will be:
svn co https://svn.sourceforge.net/svnroot/matplotlib/trunk/matplotlib

I'll post the developer checkout command on the matplotlib-devel list
when I figure it out.

Cheers!
Andrew

John Hunter wrote:
> 
> w/o addressing your original question (perhaps someone else has an
> opinion there) I suggest you wait until the 0.87.1 bug fix release
> because there is a known problem with 0.87 (images are broken).

When will this be?

-- 
Orion Poplawski
System Administrator                   303-415-9701 x222
Colorado Research Associates/NWRA      FAX: 303-415-9702
3380 Mitchell Lane, Boulder CO 80301   http://www.co-ra.com

Onsi Fakhouri had the same problem and found a workaround. But I  
appreciate your help
the WX_AGG environment variable fixed one problem and your  
information on the linking tool should prove useful
to verify that things are getting build correctly.

See my post today titled "Workaround for 0.87 WXAgg bug".
Apparently according to Onsi
  there is a bug in the C code, memory is not being allocated correctly.


On 27 Feb, 2006, at 10:24, Christopher Barker wrote:

> Sam,
>
> I wish I could be more help,. but maybe I can help steer your  
> efforts  bit:
>
>> /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/ 
>> site-packages/matplotlib/backends/backend_wxagg.py in draw(self,  
>> repaint)
>>      60         FigureCanvasAgg.draw(self)
>>      61
>> ---> 62         self.bitmap = _convert_agg_to_wx_bitmap 
>> (self.get_renderer(), None)
>>      63         if repaint:
>>      64             self.gui_repaint()
>> MemoryError: _wxagg.convert_agg_to_wx_bitmap(): could not create  
>> the wx.Bitmap
>
> This sure looks like your error occurs when MPL is trying to  
> convert from the wxagg internal image storage to a wxBitmap. As a  
> test, I'd make sure the pure Agg back-end works fine (have it  
> create a PNG). If so then font,s etc have nothing to do withyour  
> trouble...it's all about the agg -> wx conversion. that's why I  
> thought it could be caused by liniking to the wrong version of wx.
>
> to make sure you have linked to the right version, you can use:
>
> otool -L
>
> on the wxagg libs that MPL builds. That should tell you which wx  
> libs you've actuallylinked to.
>
>
> With any luck, I'll have a chance to try to get this working on my  
> system this week, and then maybe I'll be more help.
>
> -Chris
>
>
>
>
>
>
> -- 
> Christopher Barker, Ph.D.
> Oceanographer
>                                     		
> NOAA/OR&R/HAZMAT         (206) 526-6959   voice
> 7600 Sand Point Way NE   (206) 526-6329   fax
> Seattle, WA  98115       (206) 526-6317   main reception
>
> Chr...@no...

**********************************************************************
Samuel M. Smith Ph.D.
2966 Fort Hill Road
Eagle Mountain, Utah 84043
801-768-2768 voice
801-768-2769 fax
**********************************************************************
"The greatest source of failure and unhappiness in the world is
giving up what we want most for what we want at the moment"
**********************************************************************

> On Sat, 25 Feb 2006, Ryan Krauss wrote:

>> Can anyone recommend a python package for doing this kind of animation?

If you want to use wxPython, then wx.lib.floatcanvas would make this 
pretty easy.

-Chris



-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...

Sam,

I wish I could be more help,. but maybe I can help steer your efforts  bit:

> /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/backends/backend_wxagg.py 
> in draw(self, repaint)
>      60         FigureCanvasAgg.draw(self)
>      61
> ---> 62         self.bitmap = 
> _convert_agg_to_wx_bitmap(self.get_renderer(), None)
>      63         if repaint:
>      64             self.gui_repaint()
> 
> MemoryError: _wxagg.convert_agg_to_wx_bitmap(): could not create the 
> wx.Bitmap

This sure looks like your error occurs when MPL is trying to convert 
from the wxagg internal image storage to a wxBitmap. As a test, I'd make 
sure the pure Agg back-end works fine (have it create a PNG). If so then 
font,s etc have nothing to do withyour trouble...it's all about the agg 
-> wx conversion. that's why I thought it could be caused by liniking to 
the wrong version of wx.

to make sure you have linked to the right version, you can use:

otool -L

on the wxagg libs that MPL builds. That should tell you which wx libs 
you've actuallylinked to.


With any luck, I'll have a chance to try to get this working on my 
system this week, and then maybe I'll be more help.

-Chris






-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...

I thought for a bit longer and found something that works just right:

#!/usr/bin/env python

import pygtk
pygtk.require('2.0')
import gtk
import matplotlib
matplotlib.use('GTK')
from matplotlib.figure import Figure
from matplotlib.backends.backend_gtk import FigureCanvasGTK, NavigationToolbar

from Numeric import *

def close_app(widget, data=None):
    """ A call back if the close window control is clicked."""

    print "delete_event being handled" 
    gtk.main_quit() # gracefully end the application.

if __name__ == "__main__": 
    main_window = gtk.Window(gtk.WINDOW_TOPLEVEL)

    # Connect the delete_event signal to a function to close the application.
    main_window.connect( "delete_event", close_app)
    main_window.show() 


    figure = Figure(figsize=(6,4), dpi=72) 
    axes = figure.add_subplot(111)
    # This calls the imshow routine and builds an Axes instance for us.
    axes.imshow( array([[2,4,5,5],[5,6,7,8],[7,6,5,7]]))

    main_window.set_border_width(10)
   
    figure_canvas = FigureCanvasGTK(figure)
    nav_tool = NavigationToolbar(figure_canvas,main_window)
    nav_tool.show()
    figure_canvas.show()
    
    vbox = gtk.VBox()
    vbox.show()
    vbox.pack_start(figure_canvas)
    vbox.pack_start(nav_tool, False, False)
    
    main_window.add(vbox) 

    gtk.main() 

    



On Mon, Feb 27, 2006 at 11:47:57AM -0500, Jon Fox wrote:
> I'm trying to embed functionality similar to the pylab.imshow() feature
> in a gtk application, but my bony head can't find an example program where someone 
> does this explicitly, and hunting in the pylab interface has been rather
> frustrating. 
> 
> I'm using the the more typical x-y plots in gtk apps aleady (thank you
> example writers), but the ability to show an colormapped 2d image with a
> color bar would be really great.
> 
> Any takers?
> 
> -- Jon
> 
> -- 
> .*. Dr. Jon R. Fox
> ..* http://www.drfox.com
> *** jo...@dr...
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by xPML, a groundbreaking scripting language
> that extends applications into web and mobile media. Attend the live webcast
> and join the prime developer group breaking into this new coding territory!
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

-- 
.*. Dr. Jon R. Fox
..* http://www.drfox.com
*** jo...@dr...

Onsi,

Thank you very much

This work around worked for me. I haven't tested everything but I can  
make a plot successfully now with WXAgg.

Have you reported this bug?

Sam



On 26 Feb, 2006, at 17:31, Onsi Fakhouri wrote:

> Hi Samuel,
>
> I was looking through the matplotlib-users mailing list archive and  
> came across your 2006-02-25 post.  I've been having the same  
> problem and after prodding around, seem to have come up with a  
> temporary workaround.
>
> If you look in the
> /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/ 
> site-packages/matplotlib/backends/backend_wxagg.py
>
> file you'll see that there are two methods that can be used to run  
> convert_agg_to_wx_bitmap.  By default matplotlib tries to import  
> _wxagg and use the _wxagg.convert_agg_to_wx_bitmap version of this  
> method.
>
> This version is a compiled C++ shared library.  The C++ code is a  
> little cryptic, but it basically looks like the code is failing to  
> allocate memory for the bitmap object.
>
> Fortunately, backend_wxagg.py includes a python alternative called  
> _py_convert_agg_to_wx_bitmap.  This uses python code to perform the  
> agg->wx.bitmap conversion which, being written in python and not in  
> C++, is slower (the speed's fine in my opinion) -- but actually works.
>
> So for, now, a temporary solution until the C++ issue is resolved  
> is to get backend_wxagg.py to use the python versions of the  
> convert method.  To do this replace the following code in  
> backend_wxagg.py at line 228:
>
> # try to load the WXAgg accelerator
>
> try:
>     import _wxagg
> except ImportError:
>     _wxagg = None
>
> # if it's present, use it
> _use_accelerator(True)
>
> with this:
>
> # try to load the WXAgg accelerator
> try:
>     import _wxagg
> except ImportError:
>     _wxagg = None
> #It imported without any problems --- but it doesn't actually work!
> _wxagg=None
>
> # if it's present, use it
> _use_accelerator(True)
>
>
> Let me know if this helps.  It seems to have done the trick for  
> me.  But I haven't tested it too extensively.
>
> Take care,
>
> Onsi

**********************************************************************
Samuel M. Smith Ph.D.
2966 Fort Hill Road
Eagle Mountain, Utah 84043
801-768-2768 voice
801-768-2769 fax
**********************************************************************
"The greatest source of failure and unhappiness in the world is
giving up what we want most for what we want at the moment"
**********************************************************************

I'm trying to embed functionality similar to the pylab.imshow() feature
in a gtk application, but my bony head can't find an example program where someone 
does this explicitly, and hunting in the pylab interface has been rather
frustrating. 

I'm using the the more typical x-y plots in gtk apps aleady (thank you
example writers), but the ability to show an colormapped 2d image with a
color bar would be really great.

Any takers?

-- Jon

-- 
.*. Dr. Jon R. Fox
..* http://www.drfox.com
*** jo...@dr...

>>>>> "orion" == orion  <or...@co...> writes:

    orion> I'm the maintainer for python-matplotlib in Fedora Extras.
    orion> I'm getting ready to build 0.87 and am looking for
    orion> suggestions about what numeric engine to make standard.
    orion> I'm building against numpy, Numeric, and numarry, but I
    orion> don't want to have it Require: all three at run time.  Is
    orion> numpy fully functional and the way of the future and should
    orion> I just list that?  Currently (0.86) I have python-numeric
    orion> as the required engine at runtime.

w/o addressing your original question (perhaps someone else has an
opinion there) I suggest you wait until the 0.87.1 bug fix release
because there is a known problem with 0.87 (images are broken).

JDH