matplotlib-users Mailing List for matplotlib (Page 6)

I was trying the covert my program into .exe file using py2exe.

I got this message:


File "pylab.pyc", line 1, in <module>
File "matplotlib\__init__.pyc", line 733, in <module>
File "matplotlib\__init__.pyc", line 273, in wrapper
File "matplotlib\__init__.pyc", line 360, in _get_data_path
RuntimeError: Could not find the matplotlib data files

Has anyone seen this before, how to solve this problem?

Thanks

Xiaoshan

On Nov 17, 2007 7:59 AM, John Hunter <jd...@gm...> wrote:

> On Nov 17, 2007 12:26 AM, C M <cmp...@gm...> wrote:
>
> > The font problem turned out to be simply that I didn't realize that the
> > latest version of matplotlib comes with the matplotlibrc file commented
> out
> > (I had not tried to change fonts before so hadn't noticed I couldn't).
> > Uncommented the fonts section solved it.  Thanks for the tip.
> >
> > Is there a way to disable color cycling?
>
> Well, you could explicitly provide the color to "plot" but this is
> masking a deeper problem.  You do not want to e callng "plot" on every
> window redraw event.  If you post a code snippet, we might be able to
> help.
>
> JDH
>

Yeah, I believe that was what was happening.  In trying to learn to embed
plots
in wx I started from an example ("Matplotlib figure in a wx panel") found
here <http://www.scipy.org/Matplotlib_figure_in_a_wx_panel>,
and for some reason in that example the figure is calls a draw() method only
on an idle following each resize, which ultimately calls "plot".  I'm not
sure why it
is done like that.  I instead now put the draw() method just under the
__init__,
so it is drawn once initially and resized accordingly, and it works better
now.
No color changing, and in fact there had been a problem with the font
getting
distorted after a few resizes (no idea what that was about) that is now gone
too.
Very good.

Thanks very much on both issues.
CM

On Nov 17, 2007 12:26 AM, C M <cmp...@gm...> wrote:

> The font problem turned out to be simply that I didn't realize that the
> latest version of matplotlib comes with the matplotlibrc file commented out
> (I had not tried to change fonts before so hadn't noticed I couldn't).
> Uncommented the fonts section solved it.  Thanks for the tip.
>
> Is there a way to disable color cycling?

Well, you could explicitly provide the color to "plot" but this is
masking a deeper problem.  You do not want to e callng "plot" on every
window redraw event.  If you post a code snippet, we might be able to
help.

JDH

On Nov 16, 2007 9:23 AM, John Hunter <jd...@gm...> wrote:

> s
>
> On Nov 15, 2007 10:42 PM, C M <cmp...@gm...> wrote:
> > sorry, the last email did not get the pic attached.  retrying, and text
> > recopied below:
> > ----------------------------------------------
> >
> > Using Python 2.5, wxPython 2.8.4.2 (msw-unicode) matplotlib 0.90.1
> >  on winXP.  I have two questions:
> >
> >  1) I have a small working app that produces a matplotlib plot.
> >  This is a plot embedded in wxPython (it's not using Pylab).
> > When it is run as a python script, it looks fine.  However, after
> > packaging it into an .exe with py2exe (using GUI2Exe to do that),
> > the font and size of the axes are different:  smaller, and now in
> > italics. See attached picture.  I also get the errors:
> >
> > C:\mycode\dist\myapp.exe\matplotlib\__init__.py:948: UserWarning: Could
> not
> > find matplotlibrc; using defaults
> > C:\mycode\dist\myapp.exe\matplotlib\__init__.py:999: UserWarning: could
> not
> > find rc file; returning defaults
> >
> >  I did include in the data files this folder:
> >  C:\Python25\Lib\site-packages\matplotlib\mpl-data\back.png
> >
> >  and in the Python packages I included:
> >  pytz
> >  matplotlib.numerix
> >
> >  2) I thought that was the only error, but now I also see that, in
> >  either case (as a python script or as a .exe file), each time I
> >  resize the window, and the plot resizes (it is in a wxPython
> >  sizer), the color of the data line changes randomly!  It starts
> >  blue, and then as I resize it it can be green, red, whatever.
> >  In the picture attached, with resizing they've wound up different
> >  colors, though they both started blue as they should have.
>
> My guess is that you are calling plot in an event handler that is
> triggered by a resize, and are seeing mpl's color cycler.  As for the
> fonts, it looks like whatever fonts are being picked up by default
> (outside of py2exe) are not being packaged by py2exe.  Make sure all
> the fonts in site-pacakges/matplotlib/mpl-data are being picked up by
> py2exe.  In general, as long as py2exe is picking up all of mpl-data,
> I don't see why you would have trouble with fonts, your rc file, or
> your icons (eg back.png).  Of course, w/o seeing any code or setup
> files this is just a wild guess.
>
> JDH
>

The font problem turned out to be simply that I didn't realize that the
latest version of matplotlib comes with the matplotlibrc file commented out
(I had not tried to change fonts before so hadn't noticed I couldn't).
Uncommented the fonts section solved it.  Thanks for the tip.

Is there a way to disable color cycling?

cm

On Nov 16, 2007 2:54 PM, Mark Bakker <ma...@gm...> wrote:
> Erik -
>
> You write that adding a 'save' is difficult.
> I can see it is involved, but don't quite see how it is specifically difficult.
> The way I see it, every item that is added to a figure needs a method
> to write itself to a file. So you loop through all lines, polygons,
> and such, and write them to a file. I am sure you guys have thought
> about this, and maybe have a much better approach.

Last time we looked at it, there were problems pickling CXX extension
code, and in particular our transformations objects were deeply nested
CXX objects.  In the transforms branch, they will be pure python and
numpy, and we can revisit the issue again when the branch is folded
into the trunk.

Erik -

You write that adding a 'save' is difficult.
I can see it is involved, but don't quite see how it is specifically difficult.
The way I see it, every item that is added to a figure needs a method
to write itself to a file. So you loop through all lines, polygons,
and such, and write them to a file. I am sure you guys have thought
about this, and maybe have a much better approach.

Is this something that is on the to-do or wish list? Maybe not top
priority, but it would sure be nice,

Thanks,

Mark

Jordan Atlas wrote:
> Hello,
>
>     Is it possible to 'save' a matplotlib figure object using something
> like the python pickle module?  Basically, I'd like to save the 'figure'
> as a file so that I can open it and manipulate it if something is
> wrong.  Alternative suggestions are welcome.
>
> Thank you,
>
> --Jordan Atlas

Jordan,

No, this has been requested before but it is not an easy modification of
mpl to make.  The alternative suggestion is to always encapsulate the
making of a figure in a script, and then save the script and the data.
If you want to save the combination in a single file, then arrange for
the script and data to reside in a single subdirectory, and zip or tar
that subdirectory.

Eric

Hanno Klemm wrote:
> Eric,
> 
> if this is relatively straightforward, I might try and give it a shot.
> However, at the moment, I don't even have a good idea how I get pylab
> to draw an ellipsis given the two axes.
> 
> If you could point me into the right direction I might try and start
> writing some first implementation of this thing.

Hanno,

Well, I would not say it is "relatively straightforward" unless you 
understand the code in quiver, and that may take some work.  What is 
needed is very similar to the quiver code.  Or at least that is the way 
I would approach it.

Eric

> 
> Hanno
> 
> 
> Eric Firing <ef...@ha...> said:
> 
>> Michael Droettboom wrote:
>>> Don't take this is a definitive answer, because this may be out of 
>>> date...  The quiver.py module contains the following comment at
> the top:
>>> """
>>> This will also become a home for things such as standard
>>> deviation ellipses, which can and will be derived very easily from
>>> the Quiver code.
>>> """
>>>
>>> ...suggesting that this was planned but not yet implemented.
>> Yes, I never got to it, and every time I look at quiver I think, "I 
>> should get that done."  I will be needing it myself.  If no one else 
>> does it, I will get to it--but probably in weeks, not days, since I am 
>> working on other things right now.
>>
>> Eric
>>
>>> Cheers,
>>> Mike
>>>
>>> Hanno Klemm wrote:
>>>> Hi all,
>>>>
>>>> I am looking for the possibility to plot quivers with error ellipsis
>>>> on the end, representing the uncertainty in a velocity measurement. I
>>>> looked around but I haven't found a functionality for that in pylab.
>>>> Is this available or how would I go about writing such a
> functionality?
>>>> Best regards,
>>>> Hanno
>>>>
>>>>
>>>>
> 
> 
> 

Eric,

if this is relatively straightforward, I might try and give it a shot.
However, at the moment, I don't even have a good idea how I get pylab
to draw an ellipsis given the two axes.

If you could point me into the right direction I might try and start
writing some first implementation of this thing.

Hanno


Eric Firing <ef...@ha...> said:

> Michael Droettboom wrote:
> > Don't take this is a definitive answer, because this may be out of 
> > date...  The quiver.py module contains the following comment at
the top:
> > 
> > """
> > This will also become a home for things such as standard
> > deviation ellipses, which can and will be derived very easily from
> > the Quiver code.
> > """
> > 
> > ...suggesting that this was planned but not yet implemented.
> 
> Yes, I never got to it, and every time I look at quiver I think, "I 
> should get that done."  I will be needing it myself.  If no one else 
> does it, I will get to it--but probably in weeks, not days, since I am 
> working on other things right now.
> 
> Eric
> 
> > 
> > Cheers,
> > Mike
> > 
> > Hanno Klemm wrote:
> >> Hi all,
> >>
> >> I am looking for the possibility to plot quivers with error ellipsis
> >> on the end, representing the uncertainty in a velocity measurement. I
> >> looked around but I haven't found a functionality for that in pylab.
> >> Is this available or how would I go about writing such a
functionality?
> >>
> >> Best regards,
> >> Hanno
> >>
> >>
> >>
> > 
> 



-- 
Hanno Klemm
kl...@ph...

Michael Droettboom wrote:
> Don't take this is a definitive answer, because this may be out of 
> date...  The quiver.py module contains the following comment at the top:
> 
> """
> This will also become a home for things such as standard
> deviation ellipses, which can and will be derived very easily from
> the Quiver code.
> """
> 
> ...suggesting that this was planned but not yet implemented.

Yes, I never got to it, and every time I look at quiver I think, "I 
should get that done."  I will be needing it myself.  If no one else 
does it, I will get to it--but probably in weeks, not days, since I am 
working on other things right now.

Eric

> 
> Cheers,
> Mike
> 
> Hanno Klemm wrote:
>> Hi all,
>>
>> I am looking for the possibility to plot quivers with error ellipsis
>> on the end, representing the uncertainty in a velocity measurement. I
>> looked around but I haven't found a functionality for that in pylab.
>> Is this available or how would I go about writing such a functionality?
>>
>> Best regards,
>> Hanno
>>
>>
>>
> 

On Nov 16, 2007 10:03 AM, Xavier Gnata <gn...@ob...> wrote:

> /usr/lib/python2.4/site-packages/matplotlib/image.py in get_extent(self)
>     286             sz = self.get_size()
>     287             #print 'sz', sz
> --> 288             numrows, numcols = sz
>     289             if self.origin == 'upper':
>     290                 return (-0.5, numcols-0.5, numrows-0.5, -0.5)
>
> ValueError: need more than 1 value to unpack
>
> but it is another topic.


It looks like you may be passing in a 1D array to imshow or to
im.set_array.  Could you post a code sample?

JDH

Darren Dale wrote:
> On Friday 16 November 2007 10:28:25 am Xavier Gnata wrote:
>   
>> Hi,
>>
>> Quoting  matplotlib/__init__.py :
>>
>> def checkdep_ghostscript():
>>     try:
>>         if sys.platform == 'win32':
>>             command = 'gswin32c -v'
>>         else:
>>             command = 'gs -v'
>>         stdin, stdout = os.popen4(command)
>>         line = stdout.readlines()[0]
>>         v = line.split()[2]
>>         vtest = '.'.join(v.split('.')[:2]) # deal with version numbers
>> like '7.07.1'
>>         float(vtest)
>>         return vtest
>>     except (IndexError, ValueError):
>>         return None
>>
>> It fails on debian sid because 'gs -v' returns "GPL Ghostscript SVN
>> PRE-RELEASE 8.61 (2007-08-02)\n"
>>
>> Anyway, the parser will be ugly because it has to deal with version
>> numbers like '7.07.1'.
>> Should I propose a trivial patch to get thinks working on debian sid ?
>>
>> Xavier
>> ps :Why is there no standard way (like -v or --version) on *unix to get
>> the version *number*?? Only the version number. Why :(
>>     
>
>
> This was fixed a while back in svn, but thanks for the report. It turns out 
> that --version does return only the version number.
>
> Darren
>
>   
Unfortunately, I'm not able to get the svn version working :
imshow always fails :
/usr/lib/python2.4/site-packages/matplotlib/pyplot.py in imshow(*args, 
**kwargs)
   1673         hold(h)
   1674     try:
-> 1675         ret =  gca().imshow(*args, **kwargs)
   1676         draw_if_interactive()
   1677     except:

/usr/lib/python2.4/site-packages/matplotlib/axes.py in imshow(self, X, 
cmap, norm, aspect, interpolation, alpha, vmin, vmax, origin, extent, 
shape, filternorm, filterrad, imlim, **kwargs)
   4435             im.autoscale_None()
   4436
-> 4437         xmin, xmax, ymin, ymax = im.get_extent()
   4438
   4439         corners = (xmin, ymin), (xmax, ymax)

/usr/lib/python2.4/site-packages/matplotlib/image.py in get_extent(self)
    286             sz = self.get_size()
    287             #print 'sz', sz
--> 288             numrows, numcols = sz
    289             if self.origin == 'upper':
    290                 return (-0.5, numcols-0.5, numrows-0.5, -0.5)

ValueError: need more than 1 value to unpack

but it is another topic.

Xavier.


> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2005.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>   


-- 
############################################
Xavier Gnata
CRAL - Observatoire de Lyon
9, avenue Charles André
69561 Saint Genis Laval cedex
Phone: +33 4 78 86 85 28
Fax: +33 4 78 86 83 86
E-mail: gn...@ob...
############################################ 

On Friday 16 November 2007 10:28:25 am Xavier Gnata wrote:
> Hi,
>
> Quoting  matplotlib/__init__.py :
>
> def checkdep_ghostscript():
>     try:
>         if sys.platform == 'win32':
>             command = 'gswin32c -v'
>         else:
>             command = 'gs -v'
>         stdin, stdout = os.popen4(command)
>         line = stdout.readlines()[0]
>         v = line.split()[2]
>         vtest = '.'.join(v.split('.')[:2]) # deal with version numbers
> like '7.07.1'
>         float(vtest)
>         return vtest
>     except (IndexError, ValueError):
>         return None
>
> It fails on debian sid because 'gs -v' returns "GPL Ghostscript SVN
> PRE-RELEASE 8.61 (2007-08-02)\n"
>
> Anyway, the parser will be ugly because it has to deal with version
> numbers like '7.07.1'.
> Should I propose a trivial patch to get thinks working on debian sid ?
>
> Xavier
> ps :Why is there no standard way (like -v or --version) on *unix to get
> the version *number*?? Only the version number. Why :(


This was fixed a while back in svn, but thanks for the report. It turns out 
that --version does return only the version number.

Darren

Hi,

Quoting  matplotlib/__init__.py :

def checkdep_ghostscript():
    try:
        if sys.platform == 'win32':
            command = 'gswin32c -v'
        else:
            command = 'gs -v'
        stdin, stdout = os.popen4(command)
        line = stdout.readlines()[0]
        v = line.split()[2]
        vtest = '.'.join(v.split('.')[:2]) # deal with version numbers 
like '7.07.1'
        float(vtest)
        return vtest
    except (IndexError, ValueError):
        return None

It fails on debian sid because 'gs -v' returns "GPL Ghostscript SVN 
PRE-RELEASE 8.61 (2007-08-02)\n"

Anyway, the parser will be ugly because it has to deal with version 
numbers like '7.07.1'.
Should I propose a trivial patch to get thinks working on debian sid ?

Xavier
ps :Why is there no standard way (like -v or --version) on *unix to get 
the version *number*?? Only the version number. Why :(

-- 
############################################
Xavier Gnata
CRAL - Observatoire de Lyon
9, avenue Charles André
69561 Saint Genis Laval cedex
Phone: +33 4 78 86 85 28
Fax: +33 4 78 86 83 86
E-mail: gn...@ob...
############################################ 

s

On Nov 15, 2007 10:42 PM, C M <cmp...@gm...> wrote:
> sorry, the last email did not get the pic attached.  retrying, and text
> recopied below:
> ----------------------------------------------
>
> Using Python 2.5, wxPython 2.8.4.2 (msw-unicode) matplotlib 0.90.1
>  on winXP.  I have two questions:
>
>  1) I have a small working app that produces a matplotlib plot.
>  This is a plot embedded in wxPython (it's not using Pylab).
> When it is run as a python script, it looks fine.  However, after
> packaging it into an .exe with py2exe (using GUI2Exe to do that),
> the font and size of the axes are different:  smaller, and now in
> italics. See attached picture.  I also get the errors:
>
> C:\mycode\dist\myapp.exe\matplotlib\__init__.py:948: UserWarning: Could not
> find matplotlibrc; using defaults
> C:\mycode\dist\myapp.exe\matplotlib\__init__.py:999: UserWarning: could not
> find rc file; returning defaults
>
>  I did include in the data files this folder:
>  C:\Python25\Lib\site-packages\matplotlib\mpl-data\back.png
>
>  and in the Python packages I included:
>  pytz
>  matplotlib.numerix
>
>  2) I thought that was the only error, but now I also see that, in
>  either case (as a python script or as a .exe file), each time I
>  resize the window, and the plot resizes (it is in a wxPython
>  sizer), the color of the data line changes randomly!  It starts
>  blue, and then as I resize it it can be green, red, whatever.
>  In the picture attached, with resizing they've wound up different
>  colors, though they both started blue as they should have.

My guess is that you are calling plot in an event handler that is
triggered by a resize, and are seeing mpl's color cycler.  As for the
fonts, it looks like whatever fonts are being picked up by default
(outside of py2exe) are not being packaged by py2exe.  Make sure all
the fonts in site-pacakges/matplotlib/mpl-data are being picked up by
py2exe.  In general, as long as py2exe is picking up all of mpl-data,
I don't see why you would have trouble with fonts, your rc file, or
your icons (eg back.png).  Of course, w/o seeing any code or setup
files this is just a wild guess.

JDH

> Thanks for (the only) reply Eric. I guess I have to switch over to
> another plotting tool if I really want the color (maybe Gnuplot). Too
> bad I started to like matplotlibs 3D plotting.

   You can try Mayavi (it may require Mayavi2, see=20
http://scipy.org/Cookbook/MayaVi/)

=46or surface plots, see
http://scipy.org/Cookbook/MayaVi/mlab

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

Eric Firing wrote:
> Seyit Hocuk wrote:
>> Hi everybody,
>>
>> I have sent this email before today, but I wasn't a member back then. 
>> So I don't know if it was sent. Anyway I am sending it again, if 
>> anyone knows how to color the plot_surface function (along an axis 
>> perhaps), please share your knowledge.
>>
>> Thanks,
>> Seyit
>
> Seyit,
>
> I haven't seen a reply yet, so I will give the standard caution: 3-D 
> plotting in mpl is not actively supported or maintained.
>
> Eric

Thanks for (the only) reply Eric. I guess I have to switch over to 
another plotting tool if I really want the color (maybe Gnuplot). Too 
bad I started to like matplotlibs 3D plotting.

Greetz,
Seyit

Don't take this is a definitive answer, because this may be out of 
date...  The quiver.py module contains the following comment at the top:

"""
This will also become a home for things such as standard
deviation ellipses, which can and will be derived very easily from
the Quiver code.
"""

...suggesting that this was planned but not yet implemented.

Cheers,
Mike

Hanno Klemm wrote:
> Hi all,
> 
> I am looking for the possibility to plot quivers with error ellipsis
> on the end, representing the uncertainty in a velocity measurement. I
> looked around but I haven't found a functionality for that in pylab.
> Is this available or how would I go about writing such a functionality?
> 
> Best regards,
> Hanno
> 
> 
> 

-- 
Michael Droettboom
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA

Hi all,

I am looking for the possibility to plot quivers with error ellipsis
on the end, representing the uncertainty in a velocity measurement. I
looked around but I haven't found a functionality for that in pylab.
Is this available or how would I go about writing such a functionality?

Best regards,
Hanno



-- 
Hanno Klemm
kl...@ph...

Seyit Hocuk wrote:
> Hi everybody,
> 
> I have sent this email before today, but I wasn't a member back then. So 
> I don't know if it was sent. Anyway I am sending it again, if anyone 
> knows how to color the plot_surface function (along an axis perhaps), 
> please share your knowledge.
> 
> Thanks,
> Seyit

Seyit,

I haven't seen a reply yet, so I will give the standard caution: 3-D 
plotting in mpl is not actively supported or maintained.

Eric

Jordan Atlas wrote:
> Hello,
> 
>     Is it possible to 'save' a matplotlib figure object using something 
> like the python pickle module?  Basically, I'd like to save the 'figure' 
> as a file so that I can open it and manipulate it if something is 
> wrong.  Alternative suggestions are welcome.
> 
> Thank you,
> 
> --Jordan Atlas

Jordan,

No, this has been requested before but it is not an easy modification of 
mpl to make.  The alternative suggestion is to always encapsulate the 
making of a figure in a script, and then save the script and the data. 
If you want to save the combination in a single file, then arrange for 
the script and data to reside in a single subdirectory, and zip or tar 
that subdirectory.

Eric

Using Python 2.5, wxPython 2.8.4.2 (msw-unicode) matplotlib 0.90.1
on winXP.  I have two questions:

1) I have a small working app that produces a matplotlib plot.
This is a plot embedded in wxPython (it's not using Pylab).
When it is run as a python script, it looks fine.  However, after
packaging it into an .exe with py2exe (using GUI2Exe to do that),
the font and size of the axes are different:  smaller, and now in
italics. See attached picture.  I also get the errors:

C:\mycode\dist\myapp.exe\matplotlib\__init__.py:948: UserWarning: Could not
find matplotlibrc; using defaults
C:\mycode\dist\myapp.exe\matplotlib\__init__.py:999: UserWarning: could not
find rc file; returning defaults

I did include in the data files this folder:
C:\Python25\Lib\site-packages\matplotlib\mpl-data\back.png

and in the Python packages I included:
pytz
matplotlib.numerix

2) I thought that was the only error, but now I also see that, in
either case (as a python script or as a .exe file), each time I
resize the window, and the plot resizes (it is in a wxPython
sizer), the color of the data line changes randomly!  It starts
blue, and then as I resize it it can be green, red, whatever.
In the picture attached, with resizing they've wound up different
colors, though they both started blue as they should have.

Any ideas?
Thanks.

Hello,

    Is it possible to 'save' a matplotlib figure object using something 
like the python pickle module?  Basically, I'd like to save the 'figure' 
as a file so that I can open it and manipulate it if something is 
wrong.  Alternative suggestions are welcome.

Thank you,

--Jordan Atlas

> I still hope somebody more expert will discover a less ugly, more  
> general,
> and less fragile solution.

Would using macports be an acceptable/less fragile solution for you?

My matplotlib (and pretty much everything) install via macports  
survived my Tiger -> Leopard upgrade (I did an archive install) just  
fine.

Its dependency stuff requires you to intsall numpy via macports as  
well, but I think what I did was (long time ago now):

$ sudo port install py-matplotlib

To get everything up and running. At some later point I uninstalled py- 
numpy and py-matplotlib (it kept the rest of the dependencies) and  
have since been compiling them from svn.

Everything seems to be working well for me.

> Also, with plain python, you don't get the ">>>" prompt back 'till  
> you close
> the figure window, whereas with pythonw you get it back as soon as the
> figure's displayed.
> That last is important to me, because it replicates the (to me)  
> single moust
> useful thing about ipython. That means I *don't* need to install  
> ipython,
> now.

I find ipython to be indispensable for many more reasons than just  
that. For instance, I just (relatively) recently was made aware of the  
"%debug" trick that will drop you into where your last Error was  
raised if your script bombs out so that you can poke around to see  
what went wrong. So cool ... but, hey, whatever floats your boat :-)

-steve

sec6 wrote:
> 
> sent to:  mat...@li...
> 
> I'm attempting to install matplotlib under version 10.5 ("Leopard") of  
> the Macintosh operating system.
> 
> I had an old installation, via Fink, which broke when I (foolishly)  
> upgraded from version 10.4 to version 10.5.
> 
> In principle, installing matplotlib under 10.5 should be simple,  
> because Leopard already includes all of matplotlib's dependencies:
> 	Python 2.4 (and 2.5)
> 	numpy
> 	libpng
> 	zlib
> 	freetype
> 	
> In practice, I seem to be running into trouble with freetype.
> 
> ...snip...
> 
> I am going to try some monkey business along the lines of
> 	ln /usr/X11/lib/libfreetype.6.dylib /sw/lib/freetype219/lib/ 
> libfreetype.6.dylib
> and will let you know the results.  However, even if that works, it's  
> a heinous kluge, and probably fragile, too.
> 
> 

Jed's solution (bootstrap Fink from scratch) didn't work for me (Bootstrap
worked OK, but installing matplotlib  failed with "Could not resolve
inconsistent dependencies!")

However, I have now tried the "monkey business" alluded to above, and it was
successful.
#Start by launching XWindows. now:
	cd /sw/lib/freetype219/lib
	sudo ln -s /usr/X11/lib/libfreetype.6.dylib ./libfreetype.6.dylib
	<...enters password...>
	cd ~
	pythonw2.5
	>>> import pylab
#Bingo!  no error.
	>>> fig = pylab.figure()
#Empty window appears on screen.
	>>> pylab.plot([1,2,3],[1,2,3])
	[<matplotlib.lines.Line2D instance at 0x163e558>]
#Correct plot appears in window
	>>> fig.savefig('/Users/username/Desktop/temp.ps')
#Open temp.ps in a graphics program (Canvas, as it happens). Looks good.

Using pythonw, rather than python, is important.
With pythonw, pylab.figure() displays the figure, whereas with plain python,
the figure doesn't display untill you do pylab.show().
Also, with plain python, you don't get the ">>>" prompt back 'till you close
the figure window, whereas with pythonw you get it back as soon as the
figure's displayed.
That last is important to me, because it replicates the (to me) single moust
useful thing about ipython. That means I *don't* need to install ipython,
now.

I still hope somebody more expert will discover a less ugly, more general,
and less fragile solution.



-- 
View this message in context: http://www.nabble.com/Installing-under-Leopard-%28Mac-OSX-10.5%29-tf4798891.html#a13777470
Sent from the matplotlib - users mailing list archive at Nabble.com.

Thanks John,

> I think you misread the warning on hlines.
Indeed. Problem solved.

> This error typically arises when the CXX extension code is expecting a
> sequence but getting something different.  I can't reproduce it here,
> but I am running mpl svn and not 0.90.1. Are you installing from src
> or a binary package.  In any case, make sure you have a clean install
> (remove site-packages/matplotlib and if building from src your "build"
> subdirectory before reinstalling).
> 
> Can another 0.90.1 user confirm that this is a problem on that version?
I'm using Ubuntu's (7.10) standard package and since I re-installed the
entire system, that was a *clean* install (though not from source). But,
of course, it might just be that my code was inappropriate.

Christian