matplotlib-users Mailing List for matplotlib (Page 12)

On Tuesday 04 April 2006 16:32, Steve Schmerler wrote:
> Hi
>
> In the plot produced by this here the straight line isn't really
> straight and seems to "try to interpolate" the data points.
>
>
> x = linspace(0.016, 0.017, 20)
> y = 1000*x+5 + randn(len(x))/30.0
> p = polyfit(x,y,1)
> plot(x, y, 'o', x, polyval(p, x), 'r')
> savefig('image.eps')
>
> The same happens with p[0]*x+p[1] instead of polyval(p, x).
> In the corresponding .png is OK so it seems to be a problem with the
> ps-backend.

version please

I am trying to build Matplot 0.87.2 from source which I got from

http://sourceforge.net/project/showfiles.php?group_id=3D80706

By following the directions at

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

under Debian 3.1r0a for AMD64

When I type python setup.py build, I get this error

cannot find tcl/tk headers. giving up.

/usr/lib/tcl8.4 and /usr/lib/tk8.4 both exist.

According to this previous posting in response to the same problem

http://sourceforge.net/mailarchive/message.php?msg_id=3D15032585

setupext.py must be edited to make sure the libraries are found.=20
However, I tried editing it and the error did not change.  Here is the
diff between the old and new versions of setupext.py, and I'm
attaching both versions to this message.

Can anyone suggest what else I could try?  Thanks.

Gil Citro

38,39c38,39
<     'linux2' : ['/usr/local', '/usr',],
<     'linux'  : ['/usr/local', '/usr',],
---
>     'linux2' : ['/usr/local', '/usr', '/usr/lib',],
>     'linux'  : ['/usr/local', '/usr', '/usr/lib',],
312,316c312,316
<         o.tcl_lib =3D "/usr/local/lib"
<         o.tcl_inc =3D "/usr/local/include"
<         o.tk_lib =3D "/usr/local/lib"
<         o.tk_inc =3D "/usr/local/include"
<         o.tkv =3D ""
---
>         o.tcl_lib =3D "/usr/lib"
>         o.tcl_inc =3D "/usr/include"
>         o.tk_lib =3D "/usr/lib"
>         o.tk_inc =3D "/usr/include"
>         o.tkv =3D "8.4"
337a338,340
>
>

Hi

In the plot produced by this here the straight line isn't really 
straight and seems to "try to interpolate" the data points.


x = linspace(0.016, 0.017, 20)
y = 1000*x+5 + randn(len(x))/30.0
p = polyfit(x,y,1)
plot(x, y, 'o', x, polyval(p, x), 'r')
savefig('image.eps')

The same happens with p[0]*x+p[1] instead of polyval(p, x).
In the corresponding .png is OK so it seems to be a problem with the 
ps-backend.

cheers,
steve

-- 
When danger or in doubt, run in circles, scream and shout.

This isn't a bug in fancy_arrow -- this code intentionally does  
everything in data coordinates because that's the application we have  
for the arrows (we use both the location and size to display  
quantitative information).

If you want to add an option to do things in viewport coordinates, that  
would be great, but please don't break the existing functionality. I  
suggest adding viewport_coords=False in the parameter list, set to True  
for viewport. I believe there are hooks to get the transformation  
between data and viewport coordinates in mpl already, but I will not  
have time to look into this until after I finish teaching this  
semester. As John kindly pointed out, the documentation in the module  
itself is fairly extensive, although there is no external documentation  
at this point.

 From inside the script, you'll want to access things from the axis's  
object model rather than importing pylab. Hope this is useful.

Rob


On Mar 31, 2006, at 11:27 PM, Ryan Krauss wrote:

> It seems like the problem is here:
>
> cx = float(dx)/distance
> sx = float(dy)/distance
> m = array([[cx, sx],[-sx,cx]])
> matrixmultiply(coords, M)
>
> and  this same code is used in regular arrow and fancy arrow.  This
> code is rotating by and angle whose tangent is dy/dx.  That is only
> going to look right to me if the x and y ranges are the same.  I tried
> building a hack to define my own rotation angle based on scaling dx
> and dy by the axis span, but I don't know how to determine the x and y
> limits from within patches.py.  I don't seem to have access to gca()
> or ylim() from within that script.
>
> Ryan
>
> On 4/1/06, Ryan Krauss <rya...@gm...> wrote:
>> Things are going o.k. with the fancy arrow, but I am not real happy
>> with the shape of my arrow head.  It doesn't seem like the back of the
>> arrow head is perpendicular to the arrow line.  I am generating an
>> arrow with the following code:
>>
>> arrow_params={'length_includes_head':True, 'shape':'full', \
>>         'head_starts_at_zero':True}
>>
>> arrow(-6.7, 
>> -4.5,5.4,3.2,head_width=1.5,head_length=1.0,fc='k',**arrow_params)
>>
>> It seems like the problem is that my x and y axes limits are the same.
>>  If I change the axes so they are roughly square, I get a good looking
>> arrow head, but my plot isn't zoomed very well.
>>
>> Ryan
>>
>> On 4/1/06, Ryan Krauss <rya...@gm...> wrote:
>>> I guess I am o.k.  I don't know what I was doing wrong at first.  I
>>> think I was drawing the arrow off the visible part of the plot.  I
>>> seem to get pretty good results from
>>>
>>> pylab.arrow(-30,10,10,10,head_width=2,fc='k')
>>>
>>> and I can infer some of the options from this line in arrow_demo.py  
>>> (181-183):
>>>
>>> arrow(x_pos, y_pos, x_scale*length, y_scale*length, \
>>>             fc=fc, ec=ec, alpha=alpha, width=width,  
>>> head_width=head_width, \
>>>             head_length=head_length, **arrow_params)
>>>
>>> It would be great though if the docstring was more help:
>>>
>>> Help on function arrow in module matplotlib.pylab:
>>>
>>> arrow(*args, **kwargs)
>>>     Draws arrow on specified axis from (x,y) to (x+dx,y+dy).
>>>     Addition kwargs: hold = [True|False] overrides default hold state
>>>
>>> Ryan
>>>
>>> On 3/31/06, John Hunter <jdh...@ac...> wrote:
>>>>>>>>> "Ryan" == Ryan Krauss <rya...@gm...> writes:
>>>>
>>>>     Ryan> I am struggling to learn how to use the new fancy arrow.
>>>>     Ryan> Does anyone have a short example of its use or is there
>>>>     Ryan> documentation aside from arrow_demo.py?
>>>>
>>>> Hey Ryan,
>>>>
>>>> I'm CC-ing Rob Knight, who wrote the arrow code.  Perhaps you can
>>>> provide some additional detail about what you are trying to
>>>> accomplish, what you've tried, and what you don't understand.
>>>>
>>>> Rob, the arrow_demo.py example Ryan is referring to is the one you
>>>> sent with your original contribution, ie
>>>>
>>>>    
>>>> http://svn.sourceforge.net/viewcvs.cgi/matplotlib/trunk/matplotlib/ 
>>>> examples/arrow_demo.py?view=markup&rev=2188
>>>>
>>>> Thanks!
>>>> JDH
>>>>
>>>
>>
>>
>>

	
	Thanks.  I commented out the lines you suggested and MPL compiles  
and runs.  I haven't done extensive testing, but I am able to  
generate simple plots.

Kimo


On Apr 2, 2006, at 10:42 PM, matplotlib-users- 
re...@li... wrote:
>
> Message: 1
> From: Robert Hetland <he...@ta...>
> Subject: Re: [Matplotlib-users] compile problems on MacBook Pro
> Date: Sun, 2 Apr 2006 16:12:18 -0500
> To: matplotlib-users <mat...@li...>
>
>
> I had the same problem, and solved it by deleting the lines in the
> arrayobject header file that (re)defined uint and ushort.  MPL
> compiles fine after this change.  I think this is an issue with
> changing numpy, and not with the Intel Mac in general.
>
> Has anybody else run into this issue?
>
> -Rob.
>
> On Apr 1, 2006, at 8:45 AM, Kimo Johnson wrote:
>
>> Hi,
>>
>> 	I'm trying to install matplotlib on my MacBook Pro.  I installed
>> python 2.4.3 using the Universal-MacPython-2.4.3.dmg installer and
>> then installed NumPy 0.9.6 from source.  I have libfreetype and
>> libpng in /usr/local/lib, probably from my LaTeX install (using
>> MacTeX-20060314.dmg from www.ctan.org).
>>
>> 	Below are the errors I get when I run python setup.py build for
>> matplotlib.  Any help would be appreciated.
>>
>> Thanks,
>> Kimo
>>
>>
>> In file included from src/_ns_cntr.c:28:
>> /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/
>> site-packages/numpy/core/include/numpy/arrayobject.h:139: error:
>> redefinition of typedef 'ushort'
>> /usr/include/sys/types.h:85: error: previous declaration of
>> 'ushort' was here
>> /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/
>> site-packages/numpy/core/include/numpy/arrayobject.h:140: error:
>> redefinition of typedef 'uint'
>> /usr/include/sys/types.h:86: error: previous declaration of 'uint'
>> was here
>> ...(similar errors)
>> lipo: can't figure out the architecture type of: /var/tmp//
>> ccUP1QvZ.out
>> error: Command "gcc -arch ppc -arch i386 -isysroot /Developer/SDKs/
>> MacOSX10.4u.sdk -fno-strict-aliasing -Wno-long-double -no-cpp-
>> precomp -mno-fused-madd -fno-common -dynamic -DNDEBUG -g -I/Library/
>> Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-
>> packages/numpy/core/include -I/usr/local/include -I/usr/include -I.
>> -I/Library/Frameworks/Python.framework/Versions/2.4/include/
>> python2.4 -c src/_ns_cntr.c -o build/temp.macosx-10.4-fat-2.4/src/
>> _ns_cntr.o -DSCIPY=1" failed with exit status 1
>
> ----
> Rob Hetland, Assistant Professor
> Dept of Oceanography, Texas A&M University
> p: 979-458-0096, f: 979-845-6331
> e: he...@ta..., w: http://pong.tamu.edu
>

Le Mardi 04 Avril 2006 12:27, Marin Manuel a =E9crit=A0:
> Hi,
>
> >So how to put "\f" in a variable in python?
>
> You either have to escape the \ :
>      ylab=3D"$RRR [\\frac{R(300K)}{R(10K)}]$"
>
> or use raw strings:
>      ylab=3Dr"$RRR [\frac{R(300K)}{R(10K)}]$"
>
> Marin

Ok thank you, I didn't have already integrated the concept of raw string! B=
ut=20
now it is.

Emmanuel

Hi folks,

I'm just checking in with the solution to my Panther bus error
issue, in case it might help someone else, but also to publicly
thank John for walking me through the diagnosis.

It turns out the problem was due to using an old version of gcc.
The owner of the PowerBook I was installing mpl on hadn't
installed Apple's gcc updaters.  He had gcc 3.3, but not Apple's
most recent build, which fixes some C++ issues.  As a result,
C and Fortran extensions built fine, but C++ extensions built
with pycxx resulted in bus errors.  For example, "import pylab"
produced a bus error, but 

>>> import matplotlib.backends._tkagg
>>> import matplotlib.agg

produced no errors, because these extensions are not built
with pycxx.  A rebuild after updating gcc led to a working mpl.

John figured this out from the pattern of which modules loaded
and which led to errors.  I never would've figured it out on
my own.

numpy and scipy both installed fine with the older gcc and passed
all tests.  I wonder if they should be reinstalled.  If anyone
knows about this, please let me know.  The owner is uncomfortable
installing things from source, and I only have infrequent contact
with him, so if these need fixing I'd like to do it soon.

Cheers,
Tom


-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/

Hi,

>So how to put "\f" in a variable in python?
>
You either have to escape the \ :
     ylab=3D"$RRR [\\frac{R(300K)}{R(10K)}]$"

or use raw strings:
     ylab=3Dr"$RRR [\frac{R(300K)}{R(10K)}]$"

Marin

--=20
Laboratory of Neurophysics and Physiology
UMR 8119 CNRS-Universit=E9 Ren=E9 Descartes
45, rue des Saints-P=E8res
75270 Paris Cedex 06
France

Hi,

I'd advise to use 'raw strings' like:
ylab = r'$RRR [\frac{R(300K)}{R(10K)}]$'
If you then type 'ylab' in interative mode, you'll see:
'$RRR [\\frac{R(300K)}{R(10K)}]$'
Voila: Backslashes get escaped automatically.

Cheers
Christian

On Tuesday 04 April 2006 17:24, manouchk wrote:
> Hi,
>
> My little problem is simple. I want to use a variable to define labels on a
> graph. If I want to create a fraction, I have to use function \frac but
> when I do that :
> ylab="$RRR [\frac{R(300K)}{R(10K)}]$"
> The value stored in ylab is :
> '$RRR [\x0crac{R(300K)}{R(10K)}]$'
>
> So how to put "\f" in a variable in python?
>
>

Hi,

My little problem is simple. I want to use a variable to define labels on a 
graph. If I want to create a fraction, I have to use function \frac but when 
I do that :
ylab="$RRR [\frac{R(300K)}{R(10K)}]$"
The value stored in ylab is :
'$RRR [\x0crac{R(300K)}{R(10K)}]$'

So how to put "\f" in a variable in python?

Hi all,
I'm new in the matplotlib word, and I try to plot a windrose, but I don't find 
any function doing that. Then, trying to solve that by the polar plot, I 
don't find how to say matplotlib to plot bar char on polar axes. Can someone 
gives me a way?
thanks

-- 
Lionel Roubeyrie - lro...@li...
LIMAIR
http://www.limair.asso.fr

Hi,

While the savefig() command works fine, I am unable to get anything to 
appear when I use show(). Instead, the program hangs (until I hit ctrl-C 
to exit).

Here is the verbose output of a simple script, jnk.py:

from pylab import *
plot([1,2,3])
show()

When run from Cygwin, I get the following (below). Similar output is 
produced when I try other backends. Any advice?

Thanks,

James


$ python jnk.py --verbose-helpful
matplotlib data path c:\python23\share\matplotlib
$HOME=C:\cygwin\home\James
CONFIGDIR=C:\cygwin\home\James\.matplotlib
loaded rc file c:\python23\share\matplotlib\matplotlibrc
matplotlib version 0.83.2
verbose.level helpful
interactive is False
platform is win32
numerix Numeric 23.3
numerix Numeric 23.3
font search path ['c:\\python23\\share\\matplotlib']
loaded ttfcache file C:\cygwin\home\James\.matplotlib\ttffont.cache
font search path ['c:\\python23\\share\\matplotlib']
loaded ttfcache file C:\cygwin\home\James\.matplotlib\ttffont.cache
backend TkAgg version 8.4
Traceback (most recent call last):
 File "jnk.py", line 6, in ?
   show()
 File 
"C:\PYTHON23\Lib\site-packages\matplotlib\backends\backend_tkagg.py", lin
e 75, in show
   Tk.mainloop()
 File "C:\PYTHON23\lib\lib-tk\Tkinter.py", line 284, in mainloop
   _default_root.tk.mainloop(n)
KeyboardInterrupt

Oops, my bad.

That script was just a quick sample I rigged up. The zeros weren't
actually causing the problem. It did make me look elsewhere though.

The problem was caused by me assuming I need to pass in the xticks
already logarithmically scaled, which was incorrect.

Thanks for the help Darren.

Cheers,
Josh

On Monday 03 April 2006 10:57 pm, Josh Marshall wrote:
> When I run the following script:
>
> import numpy as N
> from pylab import *
> t = N.arange[0,8*N.pi, 0.1]
> y = N.sin(t)
>
> figure()
> plot(t,y)
> xticks(N.arange(0,8*N.pi, N.pi))
>
> figure()
> semilogx(t,y)
> xticks(N.arange(0,8*N.pi, N.pi))
>
> The first plot will work fine, and set the x-ticks to be multiples of
> pi, as desired.
>
> However, the second plot does not work. I have verified this is the
> case whenever log axes are used: you can't then set the x-ticks or
> labels.
>
> Is this a bug or a design feature? 

It is neither. log(0) is not plottable, you need to pick more appropriate 
xticks.

When I run the following script:

import numpy as N
from pylab import *
t =3D N.arange[0,8*N.pi, 0.1]
y =3D N.sin(t)

figure()
plot(t,y)
xticks(N.arange(0,8*N.pi, N.pi))

figure()
semilogx(t,y)
xticks(N.arange(0,8*N.pi, N.pi))

The first plot will work fine, and set the x-ticks to be multiples of
pi, as desired.

However, the second plot does not work. I have verified this is the
case whenever log axes are used: you can't then set the x-ticks or
labels.

Is this a bug or a design feature? I am interested in doing this since
I am plotting FFTs on a log-log plot, and wish to label the x-axis
('100 kHz', '1 MHz', etc...) and the y-axis in dB.

If it helps, there are two errors that are given. The first is when
attempting calling the xticks function, which throws an exception:

C:\Python24\lib\site-packages\matplotlib\figure.py in draw(self, renderer)
    527
    528         # render the axes
--> 529         for a in self.axes: a.draw(renderer)
    530
    531         # render the figure text

C:\Python24\lib\site-packages\matplotlib\axes.py in draw(self, renderer, in=
fram
)
   1407         renderer.open_group('axes')
   1408
-> 1409         try: self.transData.freeze()  # eval the lazy objects
   1410         except ValueError:
   1411             print >> sys.stderr, 'data freeze value error', self.ge=
t_po
ition(), self.dataLim.get_bounds(), self.viewLim.get_bounds()

ValueError: Domain error on eval_scalars in Transformation::freeze

The second is then when the mouse is moved over the plot, or we try to
update the plot, and the exceptions are:

  File "C:\Python24\lib\site-packages\matplotlib\axes.py", line 633,
in format_coord
    xs =3D self.format_xdata(x)
  File "C:\Python24\lib\site-packages\matplotlib\axes.py", line 615,
in format_xdata
    val =3D func(x)
  File "C:\Python24\lib\site-packages\matplotlib\ticker.py", line 401,
in format_data
    value =3D strip_math(self.__call__(value))
  File "C:\Python24\lib\site-packages\matplotlib\ticker.py", line 471,
in __call__
    fx =3D math.log(x)/math.log(b)
TypeError: a float is required

Thanks,
Josh

> > Cocoa-agg works and has been there for the last few releases.
>
> Will it build by default on OS-X?

It is pure python and a bundled nib, so it is included in all distributions=
.

I think 1.4g is the way to go. It's fixed in svn 2252.

On Sunday 02 April 2006 22:55, John Hunter wrote:
> >>>>> "Steve" == Steve Schmerler <el...@gm...> writes:
>
>     Steve> I get strange "histogram-like" plots if the y-axes scale is
>     Steve> some orders of magnitude smaller than the x-axis (latest
>     Steve> svn version).
>
> This appears to be a truncation error due to the formating string we
> use to represent floats in backend_ps.  Darren, should we prefer 1.3e
> instead of 1.3f, or perhaps 1.4e or 1.5e?
>
> JDH
>
>
>
> -------------------------------------------------------
> 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

-- 
Darren S. Dale, Ph.D.
Cornell High Energy Synchrotron Source
Cornell University
200L Wilson Lab
Rt. 366 & Pine Tree Road
Ithaca, NY 14853

dd...@co...
office: (607) 255-9894
fax: (607) 255-9001

John Hunter wrote:
>>>>>>"Steve" == Steve Schmerler <el...@gm...> writes:
> 
> 
>     Steve> I get strange "histogram-like" plots if the y-axes scale is
>     Steve> some orders of magnitude smaller than the x-axis (latest
>     Steve> svn version).
> 
> This appears to be a truncation error due to the formating string we
> use to represent floats in backend_ps.  Darren, should we prefer 1.3e
> instead of 1.3f, or perhaps 1.4e or 1.5e?
> 

Replacing %f by %e and %1.3f by %1.5e in the latest svn version's 
backend_ps.py solves the problem.

cheers,
steve

-- 
Random number generation is the art of producing pure gibberish as 
quickly as possible.
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Diwaker Gupta wrote:
> Re-sending one last time in case someone has any ideas.
> 
> ============
> In gnuplot, as one plots multiple lines on the same graph, the default
> behavior is that gnuplot automatically selects linestyle, colors and
> markers. In matplotlib, the default behavior (if I don't specify
> format etc) is to draw every line in solid style, no markers, blue
> color. I much prefer gnuplot's default behavior -- is there some way
> of configuring matplotlib to do the same?
> 
> Thanks,
> Diwaker

The short answer is no, this is not configurable.  You can get somewhat 
the same effect with some programming.  Here is a minimal illustration 
in ipython, using the attached module:

In [1]:from cycle_markers import cplot

In [2]:x = arange(0,10, 0.1)

In [3]:y1 = sin(x)

In [4]:y2 = cos(x)

In [5]:cplot(x, y1, x, y2)

If you want to be able to call cplot repeatedly and get the next 
color/marker combination in the cycle each time, you could use a 
generator instead of the simple markers list.

Eric

Re-sending one last time in case someone has any ideas.

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
In gnuplot, as one plots multiple lines on the same graph, the default
behavior is that gnuplot automatically selects linestyle, colors and
markers. In matplotlib, the default behavior (if I don't specify
format etc) is to draw every line in solid style, no markers, blue
color. I much prefer gnuplot's default behavior -- is there some way
of configuring matplotlib to do the same?

Thanks,
Diwaker
--
Web/Blog/Gallery: http://floatingsun.net/blog

John Hunter wrote:
>>>>>>"Steve" == Steve Schmerler <el...@gm...> writes:
> 
> 
>     Steve> I get strange "histogram-like" plots if the y-axes scale is
>     Steve> some orders of magnitude smaller than the x-axis (latest
>     Steve> svn version).
> 
> This appears to be a truncation error due to the formating string we
> use to represent floats in backend_ps.  Darren, should we prefer 1.3e
> instead of 1.3f, or perhaps 1.4e or 1.5e?

John, Darren,

I would suggest something like 1.5g, which will be less bulky for common 
cases than 1.5f.

Matlab ps files have all paths already transformed to paper coordinates 
but in very small units (thousandths of an inch?) so that they can be 
given as integers.

Eric

>>>>> "Steve" == Steve Schmerler <el...@gm...> writes:

    Steve> I get strange "histogram-like" plots if the y-axes scale is
    Steve> some orders of magnitude smaller than the x-axis (latest
    Steve> svn version).

This appears to be a truncation error due to the formating string we
use to represent floats in backend_ps.  Darren, should we prefer 1.3e
instead of 1.3f, or perhaps 1.4e or 1.5e?

JDH

I had the same problem, and solved it by deleting the lines in the  
arrayobject header file that (re)defined uint and ushort.  MPL  
compiles fine after this change.  I think this is an issue with  
changing numpy, and not with the Intel Mac in general.

Has anybody else run into this issue?

-Rob.

On Apr 1, 2006, at 8:45 AM, Kimo Johnson wrote:

> Hi,
>
> 	I'm trying to install matplotlib on my MacBook Pro.  I installed  
> python 2.4.3 using the Universal-MacPython-2.4.3.dmg installer and  
> then installed NumPy 0.9.6 from source.  I have libfreetype and  
> libpng in /usr/local/lib, probably from my LaTeX install (using  
> MacTeX-20060314.dmg from www.ctan.org).
>
> 	Below are the errors I get when I run python setup.py build for  
> matplotlib.  Any help would be appreciated.
>
> Thanks,
> Kimo
>
>
> In file included from src/_ns_cntr.c:28:
> /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/ 
> site-packages/numpy/core/include/numpy/arrayobject.h:139: error:  
> redefinition of typedef 'ushort'
> /usr/include/sys/types.h:85: error: previous declaration of  
> 'ushort' was here
> /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/ 
> site-packages/numpy/core/include/numpy/arrayobject.h:140: error:  
> redefinition of typedef 'uint'
> /usr/include/sys/types.h:86: error: previous declaration of 'uint'  
> was here
> ...(similar errors)
> lipo: can't figure out the architecture type of: /var/tmp// 
> ccUP1QvZ.out
> error: Command "gcc -arch ppc -arch i386 -isysroot /Developer/SDKs/ 
> MacOSX10.4u.sdk -fno-strict-aliasing -Wno-long-double -no-cpp- 
> precomp -mno-fused-madd -fno-common -dynamic -DNDEBUG -g -I/Library/ 
> Frameworks/Python.framework/Versions/2.4/lib/python2.4/site- 
> packages/numpy/core/include -I/usr/local/include -I/usr/include -I.  
> -I/Library/Frameworks/Python.framework/Versions/2.4/include/ 
> python2.4 -c src/_ns_cntr.c -o build/temp.macosx-10.4-fat-2.4/src/ 
> _ns_cntr.o -DSCIPY=1" failed with exit status 1

----
Rob Hetland, Assistant Professor
Dept of Oceanography, Texas A&M University
p: 979-458-0096, f: 979-845-6331
e: he...@ta..., w: http://pong.tamu.edu

I get strange "histogram-like" plots if the y-axes scale is some orders 
of magnitude smaller than the x-axis (latest svn version).

In [43]: matplotlib.__version__
Out[43]: '0.88svn'

# bad
x = arange(0,10,.1); y = sin(x)/100.0; plot(x,y); savefig('small_y.eps')

# good
x = arange(0,10,.1); y = sin(x); plot(x,y); savefig('normal_y.eps')

This doesn't happen if I export as .png

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