matplotlib-users Mailing List for matplotlib (Page 10)

Hi All,

On a client machine (XP) I get the following traceback.  I can't 
recreate this either on my Win XP or my Win 2000 machine.

Does anyone know what DLL numeric is trying to load here.

Thanks for any hints/help on this.

Best regards
Werner


Traceback (most recent call last):
   File "twcbF.pyo", line 1598, in OnToolbarChart
   File "frameplotmpl.pyo", line 20, in ?
   File "matplotlib\numerix\__init__.pyo", line 58, in ?
   File "Numeric.pyo", line 91, in ?
   File "multiarray.pyo", line 9, in ?
   File "multiarray.pyo", line 7, in __load
ImportError: DLL load failed: Le module spécifié est introuvable.

Hi John,

Any reason why you need to use matplotlib to do  this? In my view of
things it would be easier to use R itself with the RSPython package:
http://www.omegahat.org/RSPython/index.html.

This way you can get whatever you want done in python and then just
plot the results in R. This is vastly preferable as there is no
formula expressions in Python as there are in R, and that is 95% of
Trellis's power.  I guess you could cook up the formula capability
yourself, but why bother?

If you could explain a little more on what you want to do, I could
probably give the best course of action.

-Chris Neff

On 10/07/05, John Pitney <jo...@pi...> wrote:
> Hi,
>=20
> I'm a new user of matplotlib, and I'm trying to learn how to do some of
> the things I'm accustomed to doing in the R environment with matplotlib
> and numarray.  I find R's Lattice implentation of Trellis graphics very
> useful in my industrial data analysis work.
>=20
> For those unfamiliar with Trellis, here's the user manual:
> http://www.stat.purdue.edu/~wsc/papers/trellis.user.pdf
>=20
> Does anyone have any experience with making conditioned plots or
> scatterplot matrices with matplotlib?  I'd like to try them out with
> matplotlib and would appreciate seeing any code that's already out there.
>=20
> Thanks to the matplotlib developers for what looks like a great tool!
>=20
> Best regards,
> John
>=20
>=20
> -------------------------------------------------------
> This SF.Net email is sponsored by the 'Do More With Dual!' webinar happen=
ing
> July 14 at 8am PDT/11am EDT. We invite you to explore the latest in dual
> core and dual graphics technology at this free one hour event hosted by H=
P,
> AMD, and NVIDIA.  To register visit http://www.hp.com/go/dualwebinar
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Hi,

I'm a new user of matplotlib, and I'm trying to learn how to do some of 
the things I'm accustomed to doing in the R environment with matplotlib 
and numarray.  I find R's Lattice implentation of Trellis graphics very 
useful in my industrial data analysis work.

For those unfamiliar with Trellis, here's the user manual:
http://www.stat.purdue.edu/~wsc/papers/trellis.user.pdf

Does anyone have any experience with making conditioned plots or 
scatterplot matrices with matplotlib?  I'd like to try them out with 
matplotlib and would appreciate seeing any code that's already out there.

Thanks to the matplotlib developers for what looks like a great tool!

Best regards,
John

>>>>> "Ryan" == Ryan Krauss <rya...@co...> writes:

    Ryan> I just updated to 0.83 and I am now having a problem with
    Ryan> Ipython.  Running the following code as a script produces
    Ryan> the error message below.  I am running Python 2.3.5 on
    Ryan> Windows XP with Ipython 0.6.15.  I get the same message
    Ryan> regardless of whether or not I set ion().

    Ryan> How do I fix this?

Update from matplotlib CVS -- this bug was reported on Friday in the
thread "ipython and matplotlib-0.83" and will be included in a bug-fix
release next week.

http://sourceforge.net/mailarchive/message.php?msg_id=12305161

JDH

>>>>> "Donour" == Donour Sizemore <do...@uc...> writes:

    Donour> I'm having pretty serious performance problems using the
    Donour> wxpython backend. On linux, it takes about 1-2 seconds to
    Donour> render a small window with 1000 data points. On my mac
    Donour> laptop, it takes about three times as long.

    Donour> I've verified that interactive mode is off, antialiasing
    Donour> is disabled, and the machine is otherwise idle. I'm using:

    Donour> matplotlib 0.82 wxwindows 2.6.0.0 numeric 23.8

    Donour> Profiling shows that all the time is being sucked into a
    Donour> couple of axis drawing functions by way of OnPaint.

    Donour> Any idea why it's so slow?


Can you post a complete script that is slow for you?  If it is a
marker plot and you are using 'backend : WX', it is likely to be pretty
slow because of a problem with marker drawing that was fixed for the
*Agg backends.

Also, are you using wx or wxagg?  The WXAgg backend is likely faster
and certainly will produce better output with more features.

After you come up with your test script, please run it with
--verbose-helpful and post the script and the output.

Thanks,
JDH

I'm having pretty serious performance problems using the wxpython
backend. On linux, it takes about 1-2 seconds to render a small
window with 1000 data points. On my mac laptop, it takes about three
times as long.

I've verified that interactive mode is off, antialiasing is disabled,
and the machine is otherwise idle. I'm using:

matplotlib 0.82
wxwindows 2.6.0.0
numeric 23.8

Profiling shows that all the time is being sucked into a couple
of axis drawing functions by way of OnPaint.

Any idea why it's so slow?

thanks

donour

|Donour Sizemore              do...@cs...|
|Technical Programmer     &   Numerical Analyst     |
|Economics Research Center    Ph: 773-834-4399      |
|University of Chicago        Office: Walker 303-a  |

Hi John,
thanks for the help pointing to the area to look at.
I've made some progress on this. Just replying here rather than the dev 
list to keep this thread together.

On Win98, running "dvipng --version" returms

DVIPNG.EXE (dvipng) 1.1
kpathsea version 3.3.2
Copyright (C) 2002-2004 Jan-+ke Larsson.
There is NO warranty.  You may redistribute this software
under the terms of the GNU General Public License.
For more information about these matters, see the files
named COPYING and dvipng.c.

On Win2k it is lower case, like in linux.

The other problem is that Windows doesn't know about ' characters in 
command-line commands, so when calling latex and dvipng I changed "''" 
sequences to '""' to keep windows happy. I can't remember whether linux 
is happy with this, but I'm sure you'll know. You may have to treat the 
two OSes differently.

After making these changes it works in Win2k. Curiously there's 
something nasty going on in Win98 still where it gets partway through 
the tex demo and freezes saying there are system resource problems. I 
suspect this is to do with pipes not being flushed, but I'm unlikely to 
have time to look at this for a few weeks (I've got a 6 hour quantum 
field theory exam coming up - eek).
It would be worth incorporating the changes because it should get most 
Windows users going with TeX. Note also that I'm using the MiKTeX 
distribution, although I would expect other TeX distros to be fine.
In summary, to get TeX working under Win2k I made the following changes:

Changed line:
             if line.startswith('dvipng '):
to
             if line.lower().startswith('dvipng'):

Changed line:
             command = "latex -interaction=nonstopmode '%s'"%fname
to
             command = 'latex -interaction=nonstopmode "%s"'%fname

Changed line:
         command = "dvipng -bg Transparent -fg 'rgb 0.0 0.0 0.0' -D %d 
-T tight -o %s %s"% (dpi, pngfile, dvifile)
to
         command = 'dvipng -bg Transparent -fg "rgb 0.0 0.0 0.0" -D %d 
-T tight -o %s %s'% (dpi, pngfile, dvifile)


regards,
Gary

John Hunter wrote:
>>>>>>"Gary" == Gary Ruben writes:
> 
> 
>     Gary> I'm still getting a traceback with mpl0.83 when trying to
>     Gary> run tex_demo.py under Windows (both Win98 and Win2k).  Has
>     Gary> anyone got this going under Windows?  thanks, Gary R.

I upgraded to IPython 0.6.16_cvs and still have the same problem.  I get 
the same message when I run the simple_plot.py example from MPL.  I am 
running IPython with the -pylab flag.

Ryan

Ryan Krauss wrote:

> I just updated to 0.83 and I am now having a problem with Ipython.  
> Running the following code as a script produces the error message 
> below.  I am running Python 2.3.5 on Windows XP with Ipython 0.6.15.  
> I get the same message regardless of whether or not I set ion().
>
> How do I fix this?
>
> Thanks,
>
> Ryan
>
> import pylab, scipy
> x=scipy.arange(0,1,0.1)
> y=scipy.sin(2*scipy.pi*x)
>
> #pylab.ion()
> pylab.figure(1)
> pylab.plot(x,y)
>
>
> In [13]: run pylab_test.py
> --------------------------------------------------------------------------- 
>
> exceptions.AttributeError                            Traceback (most 
> recent call
> last)
>
> E:\GT\Research\Samii\modeling\TwoD\TMM_2D_scipy\<console>
>
> C:\Python23\Lib\site-packages\IPython\iplib.py in ipmagic(arg_s)
>    140     else:
>    141         magic_args = __IPYTHON__.var_expand(magic_args)
> --> 142         return fn(magic_args)
>    143
>    144 def ipalias(arg_s):
>
> C:\Python23\Lib\site-packages\IPython\Shell.py in magic_run(self, 
> parameter_s)
>    474
>    475     def magic_run(self,parameter_s=''):
> --> 476         Magic.magic_run(self,parameter_s,runner=self.mplot_exec)
>    477
>    478     # Fix the docstring so users see the original as well
>
> C:\Python23\Lib\site-packages\IPython\Magic.py in magic_run(self, 
> parameter_s, r
> unner)
>   1381                     self.shell.user_ns.update(prog_ns)
>   1382         finally:
> -> 1383             sys.argv = save_argv
>   1384         return stats
>   1385
>
> C:\Python23\Lib\site-packages\IPython\Shell.py in mplot_exec(self, 
> fname, *where
> , **kw)
>    465         self.matplotlib.interactive(isInteractive)
>    466         # make rendering call now, if the user tried to do it
> --> 467         if self.pylab.draw_if_interactive.called:
>    468             self.pylab.draw()
>    469             self.pylab.draw_if_interactive.called = False
>
> AttributeError: 'function' object has no attribute 'called'
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by the 'Do More With Dual!' webinar 
> happening
> July 14 at 8am PDT/11am EDT. We invite you to explore the latest in dual
> core and dual graphics technology at this free one hour event hosted 
> by HP,
> AMD, and NVIDIA.  To register visit http://www.hp.com/go/dualwebinar
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

I just updated to 0.83 and I am now having a problem with Ipython.  
Running the following code as a script produces the error message 
below.  I am running Python 2.3.5 on Windows XP with Ipython 0.6.15.  I 
get the same message regardless of whether or not I set ion().

How do I fix this?

Thanks,

Ryan

import pylab, scipy
x=scipy.arange(0,1,0.1)
y=scipy.sin(2*scipy.pi*x)

#pylab.ion()
pylab.figure(1)
pylab.plot(x,y)


In [13]: run pylab_test.py
---------------------------------------------------------------------------
exceptions.AttributeError                            Traceback (most 
recent call
 last)

E:\GT\Research\Samii\modeling\TwoD\TMM_2D_scipy\<console>

C:\Python23\Lib\site-packages\IPython\iplib.py in ipmagic(arg_s)
    140     else:
    141         magic_args = __IPYTHON__.var_expand(magic_args)
--> 142         return fn(magic_args)
    143
    144 def ipalias(arg_s):

C:\Python23\Lib\site-packages\IPython\Shell.py in magic_run(self, 
parameter_s)
    474
    475     def magic_run(self,parameter_s=''):
--> 476         Magic.magic_run(self,parameter_s,runner=self.mplot_exec)
    477
    478     # Fix the docstring so users see the original as well

C:\Python23\Lib\site-packages\IPython\Magic.py in magic_run(self, 
parameter_s, r
unner)
   1381                     self.shell.user_ns.update(prog_ns)
   1382         finally:
-> 1383             sys.argv = save_argv
   1384         return stats
   1385

C:\Python23\Lib\site-packages\IPython\Shell.py in mplot_exec(self, 
fname, *where
, **kw)
    465         self.matplotlib.interactive(isInteractive)
    466         # make rendering call now, if the user tried to do it
--> 467         if self.pylab.draw_if_interactive.called:
    468             self.pylab.draw()
    469             self.pylab.draw_if_interactive.called = False

AttributeError: 'function' object has no attribute 'called'

On Sat, 09 Jul 2005, John Hunter apparently wrote: 
> It's pretty trivial to iterate over your list and call axvspan 
> repeatedly.

Absolutely.
I saw this purely as a nice user convenience.
I was not even thinking about the performance
aspect, which you mention.

> Now if you have a long list of min/max values, you'll get 
> into performance issues since axvspan creates a new 
> rectangle for each min/max pair.  To handle this case, we 
> would need to use a PolygonCollection (axvspans?)  I would 
> suggest [ (xmin1, xmax2), (xmin1, xmin2), ...] as a more 
> natural data structure for the argument.

Yes, I considered that because of its naturalness
but went for conciseness.  (I also had in the back
of my mind that you might one day accept datetime.date
compatible tuples as arguments, and typing tuples of tuples
seemed error prone.)  Either is great.

> Is this a common need? 

Pretty common in macroeconomics.
E.g., highlight the recession periods for some time series.

However, given what you have said, I should add that 
I cannot think of an example with a *long* list of values.

Cheers,
Alan

>>>>> "Gary" == Gary Ruben <gr...@bi...> writes:

    Gary> I'm still getting a traceback with mpl0.83 when trying to
    Gary> run tex_demo.py under Windows (both Win98 and Win2k).  Has
    Gary> anyone got this going under Windows?  thanks, Gary R.

    Gary> This is e-TeX, Version 3.141592-2.2 (MiKTeX 2.4) entering
    Gary> extended mode ! Undefined control sequence.  <*> 'C:\cygwin
    Gary> \home\gary\.matplotlib\tex.cache\30565a8911a6bb487e3745c0ea3c8...

I don't have a tex distro on win32 and have no immediate plans to
install one, so you may need to hunt this one down yourself.  It looks
like there may be an errant newline after c:\cygwin, though this could
be an artifact of your post.  

I suggest opening up site-packages/matplotlib/texmanager.py and adding
print statements to figure out what is going on with the various dirs,
eg, printing the value of texcache and the various files and commands
that the code tries to create and run.  In particular, in make_dvi,
print out

tex, fname and command after the call to

 command = self.get_tex_command(tex, fname)

print 'COMMAND', command
print 'TEX', tex
print 'FNAME', fname

paste these commands into an editor and try and call them with 

            stdin, stdout, stderr = os.popen3(command)
            verbose.report(''.join(stdout.readlines()), 'debug-annoying')
            err = ''.join(stderr.readlines())
            if err: verbose.report(err, 'helpful')

just as texmanager does, and try and sort out what the cause of the
error is.  Let me know if you learn anything -- good luck!

Thanks!
JDH

>>>>> "Alan" == Alan G Isaac <ai...@am...> writes:

    Alan> axvspan is useful with plot_date.  Two thoughts addressing
    Alan> only convenience not functionality.

    Alan> 1. Might it make sense for the first argument to be a list?
    Alan> This would be interpreted as [xmin1,xmax1,xmin2,xmax2,...]
    Alan> allowing multiple shadings in one go.  If breaking the
    Alan> interface is too costly, perhaps allowing matched lists for
    Alan> the xmin and xmax arguments would be an alternative.

I wouldn't want to break the existing interface.  Is it really that
It's pretty trivial to iterate over your list and call axvspan
repeatedly.  Now if you have a long list of min/max values, you'll get
into performance issues since axvspan creates a new rectangle for each
min/max pair.  To handle this case, we would need to use a
PolygonCollection (axvspans?)  I would suggest [ (xmin1, xmax2),
(xmin1, xmin2), ...] as a more natural data structure for the
argument.  Is this a common need?

    Alan> 2. If I wish to shade date ranges, am I right that I now
    Alan> should produce date objects and then use date2num to get
    Alan> xmin and xmax?  Would it be useful for axvspan to directly
    Alan> accept the tuples that would be fed to date (or datetime)
    Alan> for xmin and xmax?

Well plotdate takes datenums (as you say, dates converted with
date2num).  I think it would add, not reduce, confusion if axvspan did
anything different in the case.  To make your code read better, just
add a little helper function converting whatever form of date
representation you use to a datetime and then call it

  axvspan( todatenum(xmin), todatenum(xmax) )

JDH

JDH

axvspan is useful with plot_date.

Two thoughts addressing only convenience not functionality.

1. Might it make sense for the first argument to be a list?  
   This would be interpreted as 
   [xmin1,xmax1,xmin2,xmax2,...]
   allowing multiple shadings in one go.  If breaking the
   interface is too costly, perhaps allowing matched lists
   for the xmin and xmax arguments would be an alternative.

2. If I wish to shade date ranges, am I right that I now 
   should produce date objects and then use date2num to get 
   xmin and xmax?  Would it be useful for axvspan to 
   directly accept the tuples that would be fed to date (or 
   datetime) for xmin and xmax?

Cheers,
Alan Isaac

I'm still getting a traceback with mpl0.83 when trying to run 
tex_demo.py under Windows (both Win98 and Win2k).
Has anyone got this going under Windows?
thanks,
Gary R.

Here's the dump with debug-annoying set:

loaded rc file C:\PYTHON23\share\matplotlib\matplotlibrc
matplotlib version 0.83
verbose.level debug-annoying
interactive is False
platform is win32
loaded modules: ['__future__', 'copy_reg', 'sre_compile', 'distutils', 
'locale', '_sre', '__main__', 'site', '__builtin__', 'datetime', 
'matplotlib.tempfile', 'encodings', 'os.path', 'encodings.encodings', 
'sre_constants', 'distutils.string', 'dateutil', 'matplotlib.datetime', 
'strop', '_random', 'tempfile', 'errno', 'matplotlib.warnings', 
'encodings.codecs', 'matplotlib.sys', 're', 'ntpath', 'pytz.sys', 
'UserDict', 'distutils.sysconfig', 'encodings.exceptions', 'nt', 
'pytz.sets', 'math', 'stat', 'zipimport', 'string', 'warnings', 
'encodings.types', '_codecs', 'sets', 'distutils.os', 'matplotlib', 
'encodings.cp1252', 'sys', 'pytz.tzinfo', 'pytz', 'pytz.datetime', 
'matplotlib.__future__', 'codecs', 'distutils.re', 'matplotlib.pytz', 
'types', 'matplotlib.dateutil', '_locale', 'matplotlib.os', 'thread', 
'sre', 'bisect', 'matplotlib.distutils', 'signal', 'distutils.errors', 
'random', 'linecache', 'itertools', 'time', 'exceptions', 'sre_parse', 
'pytz.bisect', 'distutils.sys', 'os']
numerix Numeric 23.7
font search path ['C:\\PYTHON23\\share\\matplotlib']
trying fontname C:\PYTHON23\share\matplotlib\cmex10.ttf
trying fontname C:\PYTHON23\share\matplotlib\cmmi10.ttf
trying fontname C:\PYTHON23\share\matplotlib\cmr10.ttf
trying fontname C:\PYTHON23\share\matplotlib\VeraMono.ttf
trying fontname C:\PYTHON23\share\matplotlib\cmsy10.ttf
trying fontname C:\PYTHON23\share\matplotlib\VeraMoBI.ttf
trying fontname C:\PYTHON23\share\matplotlib\cmtt10.ttf
trying fontname C:\PYTHON23\share\matplotlib\Vera.ttf
$HOME=C:\cygwin\home\gary
CONFIGDIR=C:\cygwin\home\gary\.matplotlib
loaded ttfcache file C:\cygwin\home\gary\.matplotlib\ttffont.cache
matplotlib data path C:\PYTHON23\share\matplotlib
backend TkAgg version 8.4
FigureCanvasAgg.draw
RendererAgg.__init__
This is e-TeX, Version 3.141592-2.2 (MiKTeX 2.4)
entering extended mode
! Undefined control sequence.
<*> 'C:\cygwin
 
\home\gary\.matplotlib\tex.cache\30565a8911a6bb487e3745c0ea3c8...

! Undefined control sequence.
<*> 'C:\cygwin\home
 
\gary\.matplotlib\tex.cache\30565a8911a6bb487e3745c0ea3c8...

! Undefined control sequence.
<*> 'C:\cygwin\home\gary
 
\.matplotlib\tex.cache\30565a8911a6bb487e3745c0ea3c8...

! I can't find file `'C:.tex'.
<to be read again>
                    \global
<*> 'C:\cygwin\home\gary\.
 
matplotlib\tex.cache\30565a8911a6bb487e3745c0ea3c8...
Please type another input file name
! Emergency stop.
<to be read again>
                    \global
<*> 'C:\cygwin\home\gary\.
 
matplotlib\tex.cache\30565a8911a6bb487e3745c0ea3c8...

No pages of output.
Transcript written on texput.log.

This is dvipng 1.1 Copyright 2002-2004 Jan-Åke Larsson

C:\cygwin\home\gary\.matplotlib\tex.cache\30565a8911a6bb487e3745c0ea3c8224.dvi: 
No such file or directory
C:\MIKTEX\MAIN\MIKTEX\BIN\DVIPNG.EXE warning: Unimplemented color 
specification ''rgb'


Exception in Tkinter callback
Traceback (most recent call last):
   File "C:\PYTHON23\lib\lib-tk\Tkinter.py", line 1345, in __call__
     return self.func(*args)
   File 
"C:\PYTHON23\Lib\site-packages\matplotlib\backends\backend_tkagg.py", 
line 148, in resize
     self.show()
   File 
"C:\PYTHON23\Lib\site-packages\matplotlib\backends\backend_tkagg.py", 
line 151, in draw
     FigureCanvasAgg.draw(self)
   File 
"C:\PYTHON23\Lib\site-packages\matplotlib\backends\backend_agg.py", line 
383, in draw
     self.figure.draw(renderer)
   File "C:\PYTHON23\Lib\site-packages\matplotlib\figure.py", line 511, 
in draw
     for a in self.axes: a.draw(renderer)
   File "C:\PYTHON23\Lib\site-packages\matplotlib\axes.py", line 1386, 
in draw
     self.xaxis.draw(renderer)
   File "C:\PYTHON23\Lib\site-packages\matplotlib\axis.py", line 549, in 
draw
     tick.draw(renderer)
   File "C:\PYTHON23\Lib\site-packages\matplotlib\axis.py", line 148, in 
draw
     if self.label1On: self.label1.draw(renderer)
   File "C:\PYTHON23\Lib\site-packages\matplotlib\text.py", line 849, in 
draw
     self._mytext.draw(renderer)
   File "C:\PYTHON23\Lib\site-packages\matplotlib\text.py", line 336, in 
draw
     bbox, info = self._get_layout(renderer)
   File "C:\PYTHON23\Lib\site-packages\matplotlib\text.py", line 181, in 
_get_layout
     w,h = renderer.get_text_width_height(
   File 
"C:\PYTHON23\Lib\site-packages\matplotlib\backends\backend_agg.py", line 
243, in get_text_width_height
     Z = self.texmanager.get_rgba(s, size, dpi, rgb)
   File "C:\PYTHON23\Lib\site-packages\matplotlib\texmanager.py", line 
296, in get_rgba
     X = readpng(pngfile)
RuntimeError: _image_module::readpng could not open PNG file 
C:\cygwin\home\gary\.matplotlib\tex.cache\30565a8911a6bb487e3745c0ea3c8224_96.png 
for reading

John Hunter wrote:
> Normally, this would have been a bugfix release, but I reorganized the
> matplotlib configuration files so thought it best to tick the version
> number.  All of the configuration files and dirs (matplotlibrc,
> tex.cache, ttffont.cache ) now reside in $HOME/.matplotlib (on windows
> C:\Documents and Settings\youracct\.matplotlib).  Please rename your
> .matplotlibrc file to matplotlibrc and put it there.
> 
> Other changes are 
> 
> 
> 2005-07-07 Added Eric's MA set_xdata Line2D fix - JDH
> 
> 2005-07-06 Made HOME/.matplotlib the new config dir where the
>            matplotlibrc file, the ttf.cache, and the tex.cache live.
>            The new default filenames in .matplotlib have no leading
>            dot and are not hidden.  Eg, the new names are matplotlibrc
>            tex.cache ttffont.cache.  This is how ipython does it so it
>            must be right.  If old files are found, a warning is issued
>            and they are moved to the new location.  Also fixed
>            texmanager to put all files, including temp files in
>            ~/.matplotlib/tex.cache, which allows you to usetex in
>            non-writable dirs.
> 
> 
> 2005-07-05 Fixed bug #1231611 in subplots adjust layout.  The problem
>            was that the text cacheing mechanism was not using the
>            transformation affine in the key. - JDH
> 
> 2005-07-05 Fixed default backend import problem when using API (SF bug
>            # 1209354 -  see API_CHANGES for more info - JDH
> 
> 2005-07-04 backend_gtk.py: require PyGTK version 2.0.0 or higher - SC
> 
> 2005-06-30 setupext.py: added numarray_inc_dirs for building against
>            numarray when not installed in standard location - ADS
> 
> 
> 2005-06-27 backend_svg.py: write figure width, height as int, not float.
>            Update to fix some of the pychecker warnings - SC
>   
> 2005-06-23 Updated examples/agg_test.py to demonstrate curved paths
>            and fills - JDH
> 
> 2005-06-21 Moved some texmanager and backend_agg tex caching to class
>            level rather than instance level - JDH
> 
> 2005-06-20 setupext.py: fix problem where _nc_backend_gdk is installed to the
>            wrong directory - SC
> 
> 2005-06-19 Added 10.4 support for CocoaAgg. - CM
> 
> 2005-06-18 Move Figure.get_width_height() to FigureCanvasBase and return
>            int instead of float. - SC
> 
> 2005-06-18 Applied Ted Drain's QtAgg patch: 1) Changed the toolbar to
>            be a horizontal bar of push buttons instead of a QToolbar
>            and updated the layout algorithms in the main window
>            accordingly.  This eliminates the ability to drag and drop
>            the toolbar and detach it from the window.  2) Updated the
>            resize algorithm in the main window to show the correct
>            size for the plot widget as requested.  This works almost
>            correctly right now.  It looks to me like the final size of
>            the widget is off by the border of the main window but I
>            haven't figured out a way to get that information yet.  We
>            could just add a small margin to the new size but that
>            seems a little hacky.  3) Changed the x/y location label to
>            be in the toolbar like the Tk backend instead of as a
>            status line at the bottom of the widget.  4) Changed the
>            toolbar pixmaps to use the ppm files instead of the png
>            files.  I noticed that the Tk backend buttons looked much
>            nicer and it uses the ppm files so I switched them.
> 
> 
> 2005-06-17 Modified the gtk backend to not queue mouse motion events.
>            This allows for live updates when dragging a slider. - CM
> 
> 2005-06-17 Added starter CocoaAgg backend.  Only works on OS 10.3 for
>            now and requires PyObjC.  (10.4 is high priority) - CM
> 
> 2005-06-17 Upgraded pyparsing and applied Paul McGuire's suggestions
>            for speeding things up.  This more than doubles the speed
>            of mathtext in my simple tests. JDH
> 
> 2005-06-16 Applied David Cooke's subplot make_key patch
> 
> 
> Downloads at http://matplotlib.sf.net
> 
> Enjoy!
> JDH
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by the 'Do More With Dual!' webinar happening
> July 14 at 8am PDT/11am EDT. We invite you to explore the latest in dual
> core and dual graphics technology at this free one hour event hosted by HP,
> AMD, and NVIDIA.  To register visit http://www.hp.com/go/dualwebinar
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 

>>>>> "Fernando" == Fernando Perez <Fer...@co...> writes:

    Fernando> OK, it seems that Nicholas tested it and it works.  Is
    Fernando> there anything you want me to do on the ipython side of
    Fernando> things to help along for the future?

Nothing special -- just let me know when you have applied the patch I
sent you a few days ago and I will update and test it out.

Thanks,
JDH

John Hunter wrote:
>>>>>>"Fernando" == Fernando Perez <Fer...@co...> writes:
> 
> 
>     Fernando> John Hunter wrote:
>     >> The goal is for backends.draw_if_interactive and
>     >> pylab.draw_if_interactive to be the same symbol (so Shell.py
>     >> will not fail) But it is clear that they are not the same
>     >> function
> 
>     Fernando> Strange.  Sorry that I can't work with CVS today, I have
>     Fernando> a few urgent things to take care of here.  Just some
>     Fernando> suggestions:
> 
> 
> It appears that the problem was ipython Shell.py calling 
> 
>   backend.draw_if_interactive = flag_calls(backend.draw_if_interactive)
> 
> which defeated my cache.  I think I have a working workaround in CVS
> now.
> 
> Nicholas can you test?
> 
> I can roll out a bug fix release next week if these changes work.
> Currently they affect all ipython users who use "run
> somepylabscript.py" under 0.83.

OK, it seems that Nicholas tested it and it works.  Is there anything you want 
me to do on the ipython side of things to help along for the future?

Cheers,

f

John Hunter wrote:
> Hi José,

  Hi John,

> I have thought this over a bit more after talking with Fernando Perez,
> a FC user, and wanted to add that in regards to the pytz and dateutils
> packages, I think it is fine to package them separately.  My main
> motivation for including them in the distro was to keep the list of
> dependencies small, and because these packages did not have win32
> friendly installers; neither issue should concern the FC4 packagers.

  I understand your motives and I agree with them. While packaging projects
for a distribution I am sure that you also understand our reasons for not
having multiple copies of the same code all over. Not just for space but
also for the ease that comes from doing the same work once, from having
bugs solved with one release instead of having to it multiple times. Also
if no code changes in a package there is no need to package it again,
increasing bandwidth consumption unnecessarily. (It is always good to be
able of dropping a green argument in the middle of an unrelated
discussion. ;-)

> I would emphasize vis-a-vis these dependencies that the packager
> should make sure the following examples run:
> 
>     peds-pc311:~/python/projects/matplotlib> ls examples/*date*
>     examples/date_demo1.py  examples/date_demo_convert.py
>     examples/date_demo2.py  examples/date_demo_rrule.py
> 
> and in general that the script examples/backend_driver.py, which runs
> a suite of examples over several image backends, should not generate
> errors.

  I will be aware of this issues, but a note on documentation would not
hurt. ;-) _ And yes, I know I am asking too much. :-)

  Also as far as I know both dateutils and pytz seem to have a slower pace
than matplotlib, so it should not be too difficult to synchronise the
packages. (Another variant of the famous last words... ;-)

> As I mentioned in a previous response, I hope to factor some of these
> dependencies into a separate, optional matplotlib-deps package to ease
> the installation burden on the users while making the setup as clean
> as possible for package maintainers.

  Thank you for your concern, to me as a packager it helps a lot to have the
collaboration of the package authors to solve the issues that show. As
users I am even more satisfied as I have an easy to update, and having a
close to upstream, of my favourites modules for python. I am serious here.

> Let me know if I can offer any more help, and thanks for your work
> promoting matplotlib!

 It is a fine piece of software and the interest from other projects is not
a random event. :-)

> JDH

-- 
José Abílio

I confirm it's working very fine for me now with the cvs matplotlib 
version, I'm not using the cvs ipython so that will work with th version 
0.6.15.

Thank you very much John.


    N.

John Hunter wrote:

>>>>>>"Fernando" == Fernando Perez <Fer...@co...> writes:
>>>>>>            
>>>>>>
>
>    Fernando> John Hunter wrote:
>    >> The goal is for backends.draw_if_interactive and
>    >> pylab.draw_if_interactive to be the same symbol (so Shell.py
>    >> will not fail) But it is clear that they are not the same
>    >> function
>
>    Fernando> Strange.  Sorry that I can't work with CVS today, I have
>    Fernando> a few urgent things to take care of here.  Just some
>    Fernando> suggestions:
>
>
>It appears that the problem was ipython Shell.py calling 
>
>  backend.draw_if_interactive = flag_calls(backend.draw_if_interactive)
>
>which defeated my cache.  I think I have a working workaround in CVS
>now.
>
>Nicholas can you test?
>
>I can roll out a bug fix release next week if these changes work.
>Currently they affect all ipython users who use "run
>somepylabscript.py" under 0.83.
>
>JDH
>
>  
>

>>>>> "Fernando" == Fernando Perez <Fer...@co...> writes:

    Fernando> John Hunter wrote:
    >> The goal is for backends.draw_if_interactive and
    >> pylab.draw_if_interactive to be the same symbol (so Shell.py
    >> will not fail) But it is clear that they are not the same
    >> function

    Fernando> Strange.  Sorry that I can't work with CVS today, I have
    Fernando> a few urgent things to take care of here.  Just some
    Fernando> suggestions:


It appears that the problem was ipython Shell.py calling 

  backend.draw_if_interactive = flag_calls(backend.draw_if_interactive)

which defeated my cache.  I think I have a working workaround in CVS
now.

Nicholas can you test?

I can roll out a bug fix release next week if these changes work.
Currently they affect all ipython users who use "run
somepylabscript.py" under 0.83.

JDH

John Hunter wrote:

> The goal is for backends.draw_if_interactive and
> pylab.draw_if_interactive to be the same symbol (so Shell.py will not
> fail)
> 
> But it is clear that they are not the same function

Strange.  Sorry that I can't work with CVS today, I have a few urgent things 
to take care of here.  Just some suggestions:

start sprinkling

print '*** object id for draw_if_interactive:',id(draw_if_interactive)

calls everywhere across that chain call.  That id() call should always give 
the exact same value, if the draw_if_interactive name is pointing to the same 
object.  At some point, this should show where a new object is being created 
(with a new id).

Sorry not to be of more help right now,

f

>>>>> "John" == John Hunter <jdh...@ac...> writes:

>>>>> "Fernando" == Fernando Perez <Fer...@co...> writes:
    Fernando> John, is this due to the recent __init__ patch in
    Fernando> matplotlib?  I haven't committed any of that, as I am in
    Fernando> the middle of the cvs->svn transition at ipython's site,
    Fernando> and without net access at home (moving), so I'm sort of
    Fernando> crippled right now.

    John> I tried to fix backends/__init__.py to work with existing
    John> ipython distributions.  But I only tested calling plot
    John> commands from ipython and not run.  The latter exposes the
    John> bug.  I think this should be fixed on the matplotlib end,
    John> since I don't want to require everyone to always have the
    John> latest ipython.  I'll take a look and see if I can figure
    John> out a workaround.


The problem appears to be that the draw_if_interactive function pylab
gets by calling pylab_setup is not the same as the
backends.draw_if_interactive.  Thus the "called" flag you set in
Shell.py

        backend.draw_if_interactive = flag_calls(backend.draw_if_interactive)

is not set on the pylab.draw_if_interactive.  I tried to fix this by
modifying backends_init.py as follows, so that both functions would be
the same (in CVS).

  def pylab_setup():
      'return new_figure_manager, draw_if_interactive and show for pylab'
      # Import the requested backend into a generic module object
      if pylabcache is not None: return pylabcache
      backend_name = 'backend_'+backend.lower()
      backend_mod = __import__('matplotlib.backends.'+backend_name,
      ..snip..
      return new_figure_manager, draw_if_interactive, show

  # a hack to keep old versions of ipython working with mpl after bug
  # fix #1209354
  pylabcache = None
  if 'IPython.Shell' in  sys.modules:
      pylabcache = pylab_setup()
      new_figure_manager, draw_if_interactive, show = pylabcache

Thus the symbols new_figure_manager, draw_if_interactive, show are
still defined at module level when pylab is imported from ipython.


pylab gets it draw_if_interactive symbol with


  from backends import pylab_setup
  new_figure_manager, draw_if_interactive, show = pylab_setup()

The goal is for backends.draw_if_interactive and
pylab.draw_if_interactive to be the same symbol (so Shell.py will not
fail)

But it is clear that they are not the same function

In [1]: import matplotlib.backends

In [2]: matplotlib.backends.draw_if_interactive.jdh = 'test'

In [3]: draw_if_interactive.jdh
------------------------------------------------------------
Traceback (most recent call last):
  File "<console>", line 1, in ?
AttributeError: 'function' object has no attribute 'jdh'



What am I missing?  I verified that pylab_setup is returning
pylabcache when run from ipython, and yet the draw_if_interactive
function is not shared??

JDH

>>>>> "Fernando" == Fernando Perez <Fer...@co...> writes:

    Fernando> John, is this due to the recent __init__ patch in
    Fernando> matplotlib?  I haven't committed any of that, as I am in
    Fernando> the middle of the cvs->svn transition at ipython's site,
    Fernando> and without net access at home (moving), so I'm sort of
    Fernando> crippled right now.

I tried to fix backends/__init__.py to work with existing ipython
distributions.  But I only tested calling plot commands from ipython
and not run.  The latter exposes the bug.  I think this should be
fixed on the matplotlib end, since I don't want to require everyone to
always have the latest ipython.  I'll take a look and see if I can
figure out a workaround.

JDH

John Hunter wrote:
>>>>>>"Humufr" == Humufr  <hu...@ya...> writes:
> 
> 
>     Humufr> I have a problem with ipython and the new version of
>     Humufr> matplotlib (I'm using a cvs version but there are no
>     Humufr> change since yesterday so I suppose that is the version
>     Humufr> 0.83).  When I did a:
> 
> I assume you are working with ipython CVS?  Try replacing this code block
> in IPython/Shell.py
> 
>         # We need to detect at runtime whether show() is called by the user.
>         # For this, we wrap it into a decorator which adds a 'called' flag.
>         backend.draw_if_interactive = flag_calls(backend.draw_if_interactive)
> 
>         # This must be imported last in the matplotlib series, after
>         # backend/interactivity choices have been made
>         try:
>             import matplotlib.pylab as pylab
>             self.pylab = pylab
>             self.pylab_name = 'pylab'
>         except ImportError:
>             import matplotlib.matlab as matlab            
>             self.pylab = matlab
>             self.pylab_name = 'matlab'
> 
> 
> with this
> 
>         # This must be imported last in the matplotlib series, after
>         # backend/interactivity choices have been made
>         try:
>             import matplotlib.pylab as pylab
>             self.pylab = pylab
>             self.pylab_name = 'pylab'
>         except ImportError:
>             import matplotlib.matlab as matlab            
>             self.pylab = matlab
>             self.pylab_name = 'matlab'
> 
>         # We need to detect at runtime whether show() is called by the user.
>         # For this, we wrap it into a decorator which adds a 'called' flag.
>         self.pylab.draw_if_interactive = flag_calls(self.pylab.draw_if_interactive)
> 
> 
> Does that help?

John, is this due to the recent __init__ patch in matplotlib?  I haven't 
committed any of that, as I am in the middle of the cvs->svn transition at 
ipython's site, and without net access at home (moving), so I'm sort of 
crippled right now.

If this is going to become a real problem for mpl users, I'll do my best to 
push out an update soon, but it may take me a few days.

Cheers,

f

>>>>> "Humufr" == Humufr  <hu...@ya...> writes:

    Humufr> I have a problem with ipython and the new version of
    Humufr> matplotlib (I'm using a cvs version but there are no
    Humufr> change since yesterday so I suppose that is the version
    Humufr> 0.83).  When I did a:

I assume you are working with ipython CVS?  Try replacing this code block
in IPython/Shell.py

        # We need to detect at runtime whether show() is called by the user.
        # For this, we wrap it into a decorator which adds a 'called' flag.
        backend.draw_if_interactive = flag_calls(backend.draw_if_interactive)

        # This must be imported last in the matplotlib series, after
        # backend/interactivity choices have been made
        try:
            import matplotlib.pylab as pylab
            self.pylab = pylab
            self.pylab_name = 'pylab'
        except ImportError:
            import matplotlib.matlab as matlab            
            self.pylab = matlab
            self.pylab_name = 'matlab'


with this

        # This must be imported last in the matplotlib series, after
        # backend/interactivity choices have been made
        try:
            import matplotlib.pylab as pylab
            self.pylab = pylab
            self.pylab_name = 'pylab'
        except ImportError:
            import matplotlib.matlab as matlab            
            self.pylab = matlab
            self.pylab_name = 'matlab'

        # We need to detect at runtime whether show() is called by the user.
        # For this, we wrap it into a decorator which adds a 'called' flag.
        self.pylab.draw_if_interactive = flag_calls(self.pylab.draw_if_interactive)


Does that help?

JDH

I have a problem with ipython and the new version of matplotlib (I'm 
using a cvs version but there are no change since yesterday so I suppose 
that is the version 0.83).
When I did a:

run simple_plot.py

it's working but I have this message (not very nice):


---------------------------------------------------------------------------
exceptions.AttributeError                            Traceback (most 
recent call last)

matplotlib/examples/<console>

/usr/local/lib/python2.4/site-packages/IPython/iplib.py in ipmagic(arg_s)
    140     else:
    141         magic_args = __IPYTHON__.var_expand(magic_args)
--> 142         return fn(magic_args)
    143
    144 def ipalias(arg_s):

/usr/local/lib/python2.4/site-packages/IPython/Shell.py in 
magic_run(self, parameter_s)
    474
    475     def magic_run(self,parameter_s=''):
--> 476         Magic.magic_run(self,parameter_s,runner=self.mplot_exec)
    477
    478     # Fix the docstring so users see the original as well

/usr/local/lib/python2.4/site-packages/IPython/Magic.py in 
magic_run(self, parameter_s, runner)
   1373
   1374                     else:
-> 1375                         
runner(filename,prog_ns,prog_ns,exit_ignore=exit_ignore)
   1376                 if opts.has_key('i'):
   1377                     self.shell.user_ns['__name__'] = __name__save

/usr/local/lib/python2.4/site-packages/IPython/Shell.py in 
mplot_exec(self, fname, *where, **kw)
    465         self.matplotlib.interactive(isInteractive)
    466         # make rendering call now, if the user tried to do it
--> 467         if self.pylab.draw_if_interactive.called:
    468             self.pylab.draw()
    469             self.pylab.draw_if_interactive.called = False

AttributeError: 'function' object has no attribute 'called'

Thanks for matplotlib and ipython.

N.

John Hunter wrote:

>Normally, this would have been a bugfix release, but I reorganized the
>matplotlib configuration files so thought it best to tick the version
>number.  All of the configuration files and dirs (matplotlibrc,
>tex.cache, ttffont.cache ) now reside in $HOME/.matplotlib (on windows
>C:\Documents and Settings\youracct\.matplotlib).  Please rename your
>.matplotlibrc file to matplotlibrc and put it there.
>
>Other changes are 
>
>
>2005-07-07 Added Eric's MA set_xdata Line2D fix - JDH
>
>2005-07-06 Made HOME/.matplotlib the new config dir where the
>           matplotlibrc file, the ttf.cache, and the tex.cache live.
>           The new default filenames in .matplotlib have no leading
>           dot and are not hidden.  Eg, the new names are matplotlibrc
>           tex.cache ttffont.cache.  This is how ipython does it so it
>           must be right.  If old files are found, a warning is issued
>           and they are moved to the new location.  Also fixed
>           texmanager to put all files, including temp files in
>           ~/.matplotlib/tex.cache, which allows you to usetex in
>           non-writable dirs.
>
>
>2005-07-05 Fixed bug #1231611 in subplots adjust layout.  The problem
>           was that the text cacheing mechanism was not using the
>           transformation affine in the key. - JDH
>
>2005-07-05 Fixed default backend import problem when using API (SF bug
>           # 1209354 -  see API_CHANGES for more info - JDH
>
>2005-07-04 backend_gtk.py: require PyGTK version 2.0.0 or higher - SC
>
>2005-06-30 setupext.py: added numarray_inc_dirs for building against
>           numarray when not installed in standard location - ADS
>
>
>2005-06-27 backend_svg.py: write figure width, height as int, not float.
>           Update to fix some of the pychecker warnings - SC
>  
>2005-06-23 Updated examples/agg_test.py to demonstrate curved paths
>           and fills - JDH
>
>2005-06-21 Moved some texmanager and backend_agg tex caching to class
>           level rather than instance level - JDH
>
>2005-06-20 setupext.py: fix problem where _nc_backend_gdk is installed to the
>           wrong directory - SC
>
>2005-06-19 Added 10.4 support for CocoaAgg. - CM
>
>2005-06-18 Move Figure.get_width_height() to FigureCanvasBase and return
>           int instead of float. - SC
>
>2005-06-18 Applied Ted Drain's QtAgg patch: 1) Changed the toolbar to
>           be a horizontal bar of push buttons instead of a QToolbar
>           and updated the layout algorithms in the main window
>           accordingly.  This eliminates the ability to drag and drop
>           the toolbar and detach it from the window.  2) Updated the
>           resize algorithm in the main window to show the correct
>           size for the plot widget as requested.  This works almost
>           correctly right now.  It looks to me like the final size of
>           the widget is off by the border of the main window but I
>           haven't figured out a way to get that information yet.  We
>           could just add a small margin to the new size but that
>           seems a little hacky.  3) Changed the x/y location label to
>           be in the toolbar like the Tk backend instead of as a
>           status line at the bottom of the widget.  4) Changed the
>           toolbar pixmaps to use the ppm files instead of the png
>           files.  I noticed that the Tk backend buttons looked much
>           nicer and it uses the ppm files so I switched them.
>
>
>2005-06-17 Modified the gtk backend to not queue mouse motion events.
>           This allows for live updates when dragging a slider. - CM
>
>2005-06-17 Added starter CocoaAgg backend.  Only works on OS 10.3 for
>           now and requires PyObjC.  (10.4 is high priority) - CM
>
>2005-06-17 Upgraded pyparsing and applied Paul McGuire's suggestions
>           for speeding things up.  This more than doubles the speed
>           of mathtext in my simple tests. JDH
>
>2005-06-16 Applied David Cooke's subplot make_key patch
>
>
>Downloads at http://matplotlib.sf.net
>
>Enjoy!
>JDH
>
>
>-------------------------------------------------------
>This SF.Net email is sponsored by the 'Do More With Dual!' webinar happening
>July 14 at 8am PDT/11am EDT. We invite you to explore the latest in dual
>core and dual graphics technology at this free one hour event hosted by HP,
>AMD, and NVIDIA.  To register visit http://www.hp.com/go/dualwebinar
>_______________________________________________
>Matplotlib-users mailing list
>Mat...@li...
>https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>  
>