matplotlib-users Mailing List for matplotlib (Page 3)

It=C5=9B working now, thanks. Just a remark why the fonts must be in mpl-da=
ta and=20
why matplotlib is not using the fonts who are in the system?=20

N.

Le jeudi 24 ao=C3=BBt 2006 13:56, vous avez =C3=A9crit=C2=A0:
> On Thursday 24 August 2006 13:42, hu...@ya... wrote:
> > 	Hello,
> >
> > I saw that mathtext2 was include in the svn, I try to test it but even
> > the demo is not working with the error message:
> >
> >
> >
> > ValueError: unrecognized symbol "\frac3"
> > WARNING: Failure executing file: <mathtext2_demo.py>
>
> The demo doesn't work out of the box. You need to install download the
> freefont fonts into your mpl-data directory, and then add the following
> before the pylab import in mathtext2_demo.py:
>
> from matplotlib import rcParams
> rcParams['mathtext2'] =3D True
> rcParams['mathtext.rm'] =3D 'FreeSerif.ttf'
> rcParams['mathtext.it'] =3D 'FreeSerifItalic.ttf'
> rcParams['mathtext.tt'] =3D 'FreeMono.ttf'
> rcParams['mathtext.mit'] =3D 'FreeSerifItalic.ttf'
> rcParams['mathtext.cal'] =3D 'FreeSansOblique.ttf'
> rcParams['mathtext.nonascii'] =3D 'FreeSerif.ttf'

On Thursday 24 August 2006 13:42, hu...@ya... wrote:
> 	Hello,
>
> I saw that mathtext2 was include in the svn, I try to test it but even the
> demo is not working with the error message:
>
>
>
> ValueError: unrecognized symbol "\frac3"
> WARNING: Failure executing file: <mathtext2_demo.py>

The demo doesn't work out of the box. You need to install download the 
freefont fonts into your mpl-data directory, and then add the following 
before the pylab import in mathtext2_demo.py:

from matplotlib import rcParams
rcParams['mathtext2'] = True
rcParams['mathtext.rm'] = 'FreeSerif.ttf'
rcParams['mathtext.it'] = 'FreeSerifItalic.ttf'
rcParams['mathtext.tt'] = 'FreeMono.ttf'
rcParams['mathtext.mit'] = 'FreeSerifItalic.ttf'
rcParams['mathtext.cal'] = 'FreeSansOblique.ttf'
rcParams['mathtext.nonascii'] = 'FreeSerif.ttf'

Hey guys I am creating an application that deals with a few graphs, but it
does other things also.

What I would like to do is to have a wx.Panel that takes some input from
the user, and then when I get the information I plot some data.  The user
picks a start and finish point on the graph and then I apply a cubic
integration function.  It then goes to antoher function that is graphed
the user picks max and mins on that data, then the two graphs are compared
to each other.

I have the implimentation done up to creating the graph, and allowing the
user to work with it.  The problem that I am having is that there are
parts in between that i need to stop showing the graph and ask for more
user input, via text boxes.

Does anyone know an easy way to go about doing this?  I am trying to use
wxmpl because it seemed to be easy but it does not seem to have the
complete functionality that I would like to use.

Thanks,
-Kenny

	Hello, 

I saw that mathtext2 was include in the svn, I try to test it but even the 
demo is not working with the error message:



ValueError: unrecognized symbol "\frac3"
WARNING: Failure executing file: <mathtext2_demo.py>


N.

On Wednesday 23 August 2006 02:19, David Goldsmith wrote:
> First, sorry for the long email.
>
> I'm having a segmentation fault problem trying to use plot_date,

Update numpy. Or matplotlib. Or  in matplotlib/dates.py, change line 155 from
remainder = x - ix
 to 
remainder = float(x) - ix

Hi,
I'm trying to make an executable using python 2.4.3 py2exe 0.6.8 Matplot
0.87.4

I've tried the setup file described in
http://starship.python.net/crew/theller/moin.cgi/MatPlotLib
but it didn't work. The error message that I've got was this below.
Is there anybody who know how to make a setup file with
matplotlib.get_py2exe_datafiles() ?

Thanks,
Jenny



C:\DDTAF\Run>DDTAF.exe s
Traceback (most recent call last):
  File "main.py", line 3, in ?
ImportError: No module named pylab

C:\DDTAF\Run>pause
Press any key to continue . . .
<mat...@li...>

-- 
HMRD/ORR/NOS/NOAA <http://response.restoration.noaa.gov/emergencyresponse/>

Here is a snapshot of svn this morning for those wanting to work with
the numpy beta.  Both builds are for python2.4 and windows.

exe: http://tinyurl.com/gf299
egg: http://tinyurl.com/fbjmg

-Charlie

On Wednesday 23 August 2006 2:52 am, Jouni K Sepp=E4nen wrote:
> [Again copying to matplotlib-users; maybe the main developers can
> comment on whether these shortcomings in the getp/setp interface
> should be fixed.]
>
> Hi Derek,
>
> > It does seem as those these settings affect the top and bottom of
> > the graph - I was wondering if it was possible to show tickmarks
> > along the bottom edge but not the top edge?

from pylab import *

a=3Daxes()
a.xaxis.tick_bottom()
show()

[Again copying to matplotlib-users; maybe the main developers can  
comment on whether these shortcomings in the getp/setp interface  
should be fixed.]

Hi Derek,

> It does seem as those these settings affect the top and bottom of  
> the graph - I was wondering if it was possible to show tickmarks  
> along the bottom edge but not the top edge?

I don't think that's directly supported. Here's a hacky way to do it:

     lines = getp(gca(), 'xticklines')
     toplines = [ l for l in lines if getp(l, 'ydata') == (1,) ]
     setp(toplines, visible=False)

How I came up with this: I knew that I wanted to make some of the  
xticklines invisible, so I looked at the list of line objects for  
clues as to what differs between them. They seem to have xdata and  
ydata properties, and ydata is (0,) for half of the lines and (1,)  
for the other half, so it looks like it is the vertical position in  
axis coordinates. (xdata seems to be in data coordinates.)

> And the other property I do not see on the list is the one that  
> shows whether a tick goes "into" the graph or just "out" - in the  
> prc file, there is a line:
>  xtick.direction      : in     # direction: in or out
> but there is no "direction' property?

You're right, there is no obvious property to control this. Here's an  
even hackier way to do this (and one that doesn't look very future- 
proof):

     for l in getp(gca(), 'xticklines'):
         setp(l, 'marker', 5-getp(l, 'marker'))

The line objects have a marker property, which is 2 for some markers  
and 3 for the others... so I guessed that one of them means upwards  
and the other downwards, and checked this guess by flipping the  
xtick.direction parameter and looking again. So subtracting the  
marker from 5 flips the direction.

I wonder how this is done in Matlab?

>     label: any string
>
> which shows me that the Yaxis has a label - in this case a
> string - but I do not see how one can set the font properties
> for the Yaxis label as it is not Text object??

I think you cannot do this with setp alone. Use the ylabel command:

     ylabel('foo bar', fontsize=18)

-- 
Jouni

First, sorry for the long email.

I'm having a segmentation fault problem trying to use plot_date, 
somewhere deep in matplotlib.  After giving up trying to figure out what 
was going wrong from within my "real" code, I wrote a "minimal" sample 
to try to reproduce (or not) the problem; here it is:

import datetime as dt
import wx
import matplotlib as mpl
import wxmpl

class PlotPanel(wxmpl.PlotPanel):
    def __init__(self, parent, *args, **kwargs):
        wxmpl.PlotPanel.__init__(self, parent, wx.ID_ANY, *args, **kwargs)
        self.set_crosshairs(False)
        self.Figure = self.get_figure()
        ax = self.Figure.add_subplot(111)
        ax.clear()
        ax.grid(False)
        zerodate = dt.datetime(2001,2,15)
        zero = mpl.dates.date2num(zerodate)
        x = range(int(zero), int(zero)+11)
        ax.plot_date(x,x) # Problem line; works fine with plot instead 
of plot_date
        self.draw()

class MainFrame(wx.Frame):
    def __init__(self, *args, **kwargs):
        wx.Frame.__init__(self, *args, **kwargs)
        self.PlotPanel = PlotPanel(self)
        self.MainBox = wx.BoxSizer(wx.VERTICAL)
        self.MainBox.Add(self.PlotPanel, 1,  wx.EXPAND )
        self.SetSizerAndFit(self.MainBox)

class App(wx.App):
    def OnInit(self):
        self.frame = MainFrame(None)
        self.SetTopWindow(self.frame)
        self.frame.Fit()
        self.frame.Show()
        return True

if __name__ == "__main__":
    print wx.__version__
    print wxmpl.__version__
    print mpl.__version__
    app = App(0)
    app.MainLoop()

This code runs fine with plot instead of plot_date (just like my 
original code with the same substitution).  But with plot_date, I get 
(please note the versions I'm using):

wx.__version__ =  2.6.3.3
wxmpl.__version__ =  1.2.7
matplotlib.__version__ =  0.87.4
Traceback (most recent call last):
  File "debugplot_date.py", line 40, in ?
    app = App(0)
  File 
"//Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/wx-2.6-mac-unicode/wx/_core.py", 
line 7700, in __init__
    self._BootstrapApp()
  File 
"//Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/wx-2.6-mac-unicode/wx/_core.py", 
line 7352, in _BootstrapApp
    return _core_.PyApp__BootstrapApp(*args, **kwargs)
  File "debugplot_date.py", line 30, in OnInit
    self.frame = MainFrame(None)
  File "debugplot_date.py", line 23, in __init__
    self.PlotPanel = PlotPanel(self)
  File "debugplot_date.py", line 18, in __init__
    self.draw()
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/wxmpl.py", 
line 1189, in draw
    else:
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/backends/backend_wxagg.py", 
line 61, in draw
    FigureCanvasAgg.draw(self)
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/backends/backend_agg.py", 
line 391, in draw
    self.figure.draw(renderer)
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/figure.py", 
line 532, in draw
    for a in self.axes: a.draw(renderer)
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/axes.py", 
line 1045, in draw
    a.draw(renderer)
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/axis.py", 
line 548, in draw
    majorLabels = [self.major.formatter(val, i) for i, val in 
enumerate(majorLocs)]
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/dates.py", 
line 369, in __call__
    return self._formatter(x, pos)
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/dates.py", 
line 247, in __call__
    dt = num2date(x, self.tz)
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/dates.py", 
line 205, in num2date
    if not iterable(x): return _from_ordinalf(x, tz)
  File 
"/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/dates.py", 
line 156, in _from_ordinalf
    hour, remainder = divmod(24*remainder, 1)
ValueError: need more than 0 values to unpack

Using winpdb, I determined that this line in dates.py get called 13 
times before, on the 14th time, it causes a "communication failure" 
between winpdb and the code.

Any ideas what's up?  Thanks in advance,

DG

PS: I'm on a Mac, running system 10.4.7 with Python 2.4.3 (#1, Apr  7 
2006, 10:54:33) [GCC 4.0.1 (Apple Computer, Inc. build 5250)] on darwin

Briefly, is there a way to get a callback in the event that the  
canvas has been redrawn (e.g. from a zoom or pan)?

matplotlib 0.86.2
python 2.4.1
wxPython 2.6.2.1
Mac OS X.4.7

I draw a circle at a certain point. I let the user drag that point to  
a new location. However, if the plot is redrawn for any reason, I  
have to redraw my circle in a new position. I can't seem to find any  
way to get notified of a redraw.

Figure.add_axobserver() isn't fully implemented as far as I can tell  
since it only notifies for a artist deletion or a call to sca().
Axes.connect() doesn't work either because there's no way to set emit  
in a class-wide scope and consequently calls made to set_xlim made by  
the toolbar do not trigger an event because they lack the emit  
keyword in their calls. Here's the relevant code from axes.py.

   def set_xlim(self, *args, **kwargs):
         # .....
         emit = popd(kwargs, 'emit', False)
         # .....
         if emit: self._send_xlim_event()
         return vmin, vmax

I've been trying to accomplish this without editing matplotlib  
package code. Is there a way I could override Axes.set_xlim() in  
order to insert "emit=True" into the arguments?

Maybe there's a better way to do what I'm doing?


Thanks,

R.

Flavio,

One more point: please do not use quiver2 directly.  Instead, use 
quiver, which will call quiver2 unless it finds a keyword argument that 
exists only in the old version of quiver.  The name quiver2 is used only 
temporarily to provide this compatibility, which will be removed in a 
future release.

Eric

Flavio Coelho wrote:
> Hi,
> 
> I am tring to figure out how to use the c argument to the quiver2 function.

Thanks Eric,

in fact it depends on if x,y are 0 or not. But I got it to work it now.

thanks,

Fl=E1vio

On 8/22/06, Eric Firing <ef...@ha...> wrote:
>
> Flavio Coelho wrote:
> > Hi,
> >
> > I am tring to figure out how to use the c argument to the quiver2
> function.
> >
> > I want to be able to color my arrows proportionally to their size.
> >
> > I have tried this and it doesn' t work as I imagined:
> >
> > c =3D sqrt((array(u)-array(x))**2+(array(v)-array(y))**2)
>
> The length of the arrow is proportional to sqrt(u**2 + v**2),
> not to sqrt((u-x)**2 + (v-y)**2).
>
> Is the reason you are not getting what you expect?
>
> Eric
>
> >
> > quiver2(x,y,u,v,c)
> >
> > the docstring on quiver2 is not clear on how c works, any help will be
> > appreciated.
>



--=20
Fl=E1vio Code=E7o Coelho
registered Linux user # 386432
---------------------------
"Laws are like sausages. It's better not to see them being made."
Otto von Bismark

Flavio Coelho wrote:
> Hi,
> 
> I am tring to figure out how to use the c argument to the quiver2 function.
> 
> I want to be able to color my arrows proportionally to their size.
> 
> I have tried this and it doesn' t work as I imagined:
> 
> c = sqrt((array(u)-array(x))**2+(array(v)-array(y))**2)

The length of the arrow is proportional to sqrt(u**2 + v**2),
not to sqrt((u-x)**2 + (v-y)**2).

Is the reason you are not getting what you expect?

Eric

> 
> quiver2(x,y,u,v,c)
> 
> the docstring on quiver2 is not clear on how c works, any help will be 
> appreciated.

Phil, Jim,

Thanks.  I need to add a suitable example for the contourf case, because 
as you have discovered it is a little different from all the other 
mappable cases: colorbar takes its cue from the contourf arguments so as 
to be consistent with the way the contourf plot was made.  There is a 
line in the colorbar docstring indicating this, although I don't recall 
when I put that line in, so it may not be in the version you are using:

If mappable is a ContourSet, its extend kwarg is included automatically.

Eric

Philip Austin wrote:
> James Boyle writes:
>  > Thanks, Eric
>  > 
>  > to
>  > 
>  > im = contourf(Zm,cmap=palette)
>  > 
>  > I do not get the over/under pointers on the end of the colorbar.
>  > 
>  > Does this feature not work with contourf - Or am I doing something 
>  > wrong.
> 
> Here's a code segment that works for me:
> 
>     vmin,vmax=(280,305)
>     norm = colors.normalize(vmin = vmin, vmax = vmax, clip = True)
>     cmap=matplotlib.cm.jet
>     cmap.set_over('salmon', 1.0)
>     cmap.set_under('black', 0.25)
>     contplot = axes.contourf(pv.time_value,
>                              pv.pressure_value,
>                              pv.var1_value,
>                              n.arange(vmin,vmax,2.),
>                              norm=norm,
>                              extend='both',
>                              cmap=cmap)
> 
> 
> I got the hint on this from the contourf docstring
> 
>         ***** New: *****
>         * extend = 'neither', 'both', 'min', 'max'
>           Unless this is 'neither' (default), contour levels are
>           automatically added to one or both ends of the range so that
>           all data are included.  These added ranges are then
>           mapped to the special colormap values which default to
>           the ends of the colormap range, but can be set via
>           Colormap.set_under() and Colormap.set_over() methods.
>           To replace clip_ends=True and V = [-100, 2, 1, 0, 1, 2, 100],
>           use extend='both' and V = [2, 1, 0, 1, 2].
> 
> 
> regards, Phil
> 
> 
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

I was a little suprised to learn that

axes.toggle_log_lineary()

overwrites user-supplied tick locators and formatters.  My naive
assumption was that it would "toggle" only the semilog property
and leave other attributes unchanged.  Perhaps this could be
mentioned in the docstring?

regards, Phil Austin

James Boyle writes:
 > Thanks, Eric
 > 
 > to
 > 
 > im = contourf(Zm,cmap=palette)
 > 
 > I do not get the over/under pointers on the end of the colorbar.
 > 
 > Does this feature not work with contourf - Or am I doing something 
 > wrong.

Here's a code segment that works for me:

    vmin,vmax=(280,305)
    norm = colors.normalize(vmin = vmin, vmax = vmax, clip = True)
    cmap=matplotlib.cm.jet
    cmap.set_over('salmon', 1.0)
    cmap.set_under('black', 0.25)
    contplot = axes.contourf(pv.time_value,
                             pv.pressure_value,
                             pv.var1_value,
                             n.arange(vmin,vmax,2.),
                             norm=norm,
                             extend='both',
                             cmap=cmap)


I got the hint on this from the contourf docstring

        ***** New: *****
        * extend = 'neither', 'both', 'min', 'max'
          Unless this is 'neither' (default), contour levels are
          automatically added to one or both ends of the range so that
          all data are included.  These added ranges are then
          mapped to the special colormap values which default to
          the ends of the colormap range, but can be set via
          Colormap.set_under() and Colormap.set_over() methods.
          To replace clip_ends=True and V = [-100, 2, 1, 0, 1, 2, 100],
          use extend='both' and V = [2, 1, 0, 1, 2].


regards, Phil

Thanks, Eric

Ok,
I ran the image_masked.py and all went OK
BUT
if I change line 34 of image_masked.py:
im = imshow(Zm, interpolation='bilinear',
     cmap=palette,
     norm = colors.normalize(vmin = -1.0, vmax = 1.0, clip = False),
     origin='lower', extent=[-3,3,-3,3])

to

im = contourf(Zm,cmap=palette)

I do not get the over/under pointers on the end of the colorbar.

Does this feature not work with contourf - Or am I doing something 
wrong.

--Jim

On Aug 21, 2006, at 5:13 PM, Eric Firing wrote:

> Jim,
>
> Look at examples/image_masked.py.  I think that is the only example 
> with pointed ends, and I think it was present as-is in 0.87.3.
>
> Eric
>
> James Boyle wrote:
>> I am using matplotlib 0.87.3
>> The documentation of colorbar in color.py seems to indicate that one 
>> should be able to create a color bar with pointed ends, the fill 
>> color of the ends corresponding to the over and under colors.
>> I have not been able to get this to work.  I have set the 
>> colormap.set_under and set_over values but no success.
>> If this does work could someone (Eric?) post an example, of how it is 
>> done. I feel that I am missing something simple.
>> Thanks for any help.
>> --Jim

Hi,

I am tring to figure out how to use the c argument to the quiver2 function.

I want to be able to color my arrows proportionally to their size.

I have tried this and it doesn' t work as I imagined:

c =3D sqrt((array(u)-array(x))**2+(array(v)-array(y))**2)

quiver2(x,y,u,v,c)

the docstring on quiver2 is not clear on how c works, any help will be
appreciated.

thanks,
--=20
Fl=E1vio Code=E7o Coelho
registered Linux user # 386432
---------------------------
"Laws are like sausages. It's better not to see them being made."
Otto von Bismark

> The colors come from the current colormap.  You can go all out and
>  make your own colormap and color based on index using your map.  I
>  don't know how to do that off the top of my head though.

The colors may come from the colormap, but how do they get assigned to
areas of the plot itself?  The most obvious way to do it is to index
by z, so areas with low z are shown with a cool color, and areas with
high z are shown with a hot color (assuming you are using a cold-hot
colormap).

As far as I can tell though, this is not the system currently used,
nor is there any coherent system used.

array.tolist() can be used as a temporary workaround.
i.e.  tack .tolist() onto the end of any numpy.array arguments to matplotlib
functions.

--bb

On 8/22/06, Sven Schreiber <sve...@gm...> wrote:
>
> Hi all,
>
> I'm using recently changed numpy features, so I have not been able to
> use matplotlib in a while, hoping for a release compatible with numpy
> 1.0beta.
>
> I guess I could also install Numeric and use that for matplotlib in the
> meantime, as a workaround. But from posts on the devel list it seems
> that a new release is imminent?
>
> It's just that I need to get something done in matplotlib until
> tomorrow, so it would be great if somebody could tell me if I should
> workaround or just wait.
>
> Btw, I tried the egg that was mentioned in a post, but it didn't work
> for me (something about "matrix not defined" or similar).
>
> Thanks for your help,
> Sven
>
>

Hi,
I have come across a couple of bugs when zooming into a plot

1) if you goto draw a box (to where you are going to zoom) it will zoom in
fine
but if while drawing the box you release the mouse button BUT continue to
move the mouse the zoom does not function (when mouse stops) and a black box
is drawn in the plotting window

2) If you just keep zooming in eventually you get to a point where the graph
just doesn't make sense.
I mean if you plot a slope and keep zooming in smaller and smaller you
eventually get to a position where you suddenly get a "triangle" shaped plot

Hi all,

I'm using recently changed numpy features, so I have not been able to
use matplotlib in a while, hoping for a release compatible with numpy
1.0beta.

I guess I could also install Numeric and use that for matplotlib in the
meantime, as a workaround. But from posts on the devel list it seems
that a new release is imminent?

It's just that I need to get something done in matplotlib until
tomorrow, so it would be great if somebody could tell me if I should
workaround or just wait.

Btw, I tried the egg that was mentioned in a post, but it didn't work
for me (something about "matrix not defined" or similar).

Thanks for your help,
Sven

"Derek Hohls" <DH...@cs...> writes:

>   In [9]: ax.set_xlim()?
>  I get
>   Object `ax.set_xlim()` not found.

You need to do ax.set_xlim? without the parentheses.

> You suggested:
> "The list you want is precisely the output of the getp command."
> But for the getp? , I get:

I meant the output of the actual getp command, not its help text. E.g.

    In [4]:recs=bar([1,2,3],[4,5,6])

    In [5]:getp(recs)
        alpha = 1.0
        animated = False
          ...
        y = 0.0
        zorder = 1

gives you the list of properties settable with setp. Similarly
getp(gca()) gives you a long list of properties, including xticklines.

> The matplotlabprc file has a clearly labelled line that 
> addresses part of this:
>  
> xtick.major.size     : 2      # major tick size in points
>  
> but of course I would like to do this in code.

I guess it isn't very obvious how to do this with setp. It is the
markersize property (which has the abbreviation ms):

    In [25]:setp(getp(gca(), 'xticklines'), 'ms', 10)

Note that here getp returns a list of objects, and setp sets the
property on every object in the list.

But if you already know what something is called in the matplotlibrc
file, you can set it programmatically:

    In [49]:rc('xtick.major', size=5, pad=20)

The rc settings do not affect existing images, so you have to make a
new plot before you see the effect:

    In [50]:figure(); bar([1,2,3],[4,5,6])

> I guess that, overall, I have been expecting matplotlib to
> have a simple "dot" notation throughout - 
>   xaxis.xtick.major.size = 2

The getp/setp methods are part of matplotlib's pylab interface, which
is designed to reproduce Matlab's "handle graphics". There is also an
OO interface, which looks like this (this is the agg_oo.py example
from the examples subdirectory):

    fig = Figure()
    canvas = FigureCanvas(fig)
    ax = fig.add_subplot(111)
    ax.plot([1,2,3])
    ax.set_title('hi mom')
    ax.grid(True)
    ax.set_xlabel('time')
    ax.set_ylabel('volts')
    canvas.print_figure('test')

I seem to recall some discussion about making it more Pythonic, e.g. 
allowing you to do

    ax.title = 'hi mom'
    ax.xlabel = 'time'

but I don't know whether it is a high priority for any of the
developers.

-- 
Jouni