matplotlib-users Mailing List for matplotlib (Page 9)

John Hunter wrote:
> On Wed, Apr 14, 2010 at 2:14 PM, Jae-Joon Lee <lee...@gm...> wrote:
> 
>> John, the relevant code to define the "colors" attribute seems to be
>> written by you. Maybe this is some matlab convention? Can you comment
>> on this?
> 
> The original color letters did come from matlab, and some of the color
> RGB choices, like the background for the subplot gray, I got using a
> color picker on my screen for maximum compatibility, but I do not
> recall if that was the case for 'm', 'c'  and 'y'.  Most likely these
> were the rgb values of the matlab colors.
> 
> The html colors were submitted later by another user, and apparently
> we never checked for consistency.   I think this is a wart, and I
> don't feel strongly about keeping it or changing it for consistency in
> the trunk ahead of 1.0 (it should remain as is on the branch).  If we
> change it on the trunk, we should change the single letter codes to
> correspond to the html full color name standards, rather than the
> other way around.

I would be cautious about making this change. At least on my laptop 
screen, with the default white axes background, the single-letter colors 
show up better than the html versions.  It makes sense to me for the 
single-letter colors to be chosen for good visibility and contrast in 
actual use, not for consistency with the html names.  In fact, I see no 
good argument for consistency in this case.

Eric


> 
> JDH
> 
> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On Wed, Apr 14, 2010 at 2:14 PM, Jae-Joon Lee <lee...@gm...> wrote:

> John, the relevant code to define the "colors" attribute seems to be
> written by you. Maybe this is some matlab convention? Can you comment
> on this?

The original color letters did come from matlab, and some of the color
RGB choices, like the background for the subplot gray, I got using a
color picker on my screen for maximum compatibility, but I do not
recall if that was the case for 'm', 'c'  and 'y'.  Most likely these
were the rgb values of the matlab colors.

The html colors were submitted later by another user, and apparently
we never checked for consistency.   I think this is a wart, and I
don't feel strongly about keeping it or changing it for consistency in
the trunk ahead of 1.0 (it should remain as is on the branch).  If we
change it on the trunk, we should change the single letter codes to
correspond to the html full color name standards, rather than the
other way around.

JDH

On Sun, Apr 11, 2010 at 12:40 PM, Peter Butterworth <bu...@gm...> wrote:
> sorry if this has been covered before, but I must say I've found the
> following quite confusing :
> color="cyan" is not in fact equivalent to color='c'
>
>
> in colors.py  :
>
> Commands which take color arguments can use several formats to specify
> the colors.  For the basic builtin colors, you can use a single letter
>
>    - b  : blue
>    - g  : green
>    - r  : red
>    - c  : cyan
>    - m  : magenta
>    - y  : yellow
>    - k  : black
>    - w  : white
>
> in ColorConverter :
>    colors = {
>        'b' : (0.0, 0.0, 1.0),
>        'g' : (0.0, 0.5, 0.0),
>        'r' : (1.0, 0.0, 0.0),
>        'c' : (0.0, 0.75, 0.75),
>        'm' : (0.75, 0, 0.75),
>        'y' : (0.75, 0.75, 0),
>        'k' : (0.0, 0.0, 0.0),
>        'w' : (1.0, 1.0, 1.0),
>        }
>
> we are told 'c' is short for cyan. Yet color="cyan" is not equivalent
> to color='c'
>   'cyan'                 : '#00FFFF'
>
> In [50]: rgb2hex((0.0, 0.75, 0.75))
> Out[50]: '#00bfbf'
>

Thank you for reporting.

It seems that it is not just "c", but the rgb values of "m" and "y"
are also different.


In [26]: cc.to_rgb("magenta")
Out[26]: (1.0, 0.0, 1.0)

In [27]: cc.to_rgb("m")
Out[27]: (0.75, 0, 0.75)



In [30]: cc.to_rgb("yellow")
Out[30]: (1.0, 1.0, 0.0)

In [31]: cc.to_rgb("y")
Out[31]: (0.75, 0.75, 0)


John, the relevant code to define the "colors" attribute seems to be
written by you. Maybe this is some matlab convention? Can you comment
on this?

Regards,

-JJ



> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Hello there,

using matplotlib 0.99, I have a problem with colorbars, which is illustrated
by the following code:

---8<-------
from matplotlib import ticker
import numpy as np

a = np.arange(676).reshape((26,26))
x = y = np.arange(26)

colorMap = mpl.cm.get_cmap('jet', 10)
colorNorm = mpl.colors.Normalize(vmin=0, vmax=400,clip=True)

plot = contourf(x,y,a,10,cmap=colorMap,norm=colorNorm, \
     locator=ticker.LinearLocator(10))
cb = colorbar(plot,orientation='horizontal',cmap=colorMap,norm=colorNorm, \
     extend='neither',spacing='uniform')
---8<-------

So I have data ranging from 0 up to say 675. I want to create a contour plot,
but the colorbar should only go up to 400, and all values higher than that
should be shown in the color of the maximum value 400, which is working 
fine.

However, the colorbar does not go from 0 to 400, but rather from 0 to 675:

---8<-------
In [52]: cb._boundaries
Out[52]:
array([ -6.75000000e-04,   7.50000000e+01,   1.50000000e+02,
          2.25000000e+02,   3.00000000e+02,   3.75000000e+02,
          4.50000000e+02,   5.25000000e+02,   6.00000000e+02,
          6.75000675e+02])
---8<-------

The colors shown are correct, but the colorbar looks ugly, because two 
thirds
of it are filled with the color of the maximum value.

According to the documentation
(http://matplotlib.sourceforge.net/api/colors_api.html#matplotlib.colors.Normalize):

    If clip is True and the given value falls outside the range, the
returned
    value will be 0 or 1, whichever is closer.

So my question is: What do I need to do so that the colorbar actually  has
the
range as specified by my colorNorm?

Thanks for your help,

Andreas.

On Sat, Apr 10, 2010 at 10:36 AM, arsbbr <ar...@gm...> wrote:
> but it doesn't
> work.
>

Can you provide more details of what you tried and "how" it does not
work. For example, an exception is raised? or code runs without any
error but the result is not correct?

With my simple test, all things seem fine. I didn't carefully check if
the output has a correct font selected, but at least the font was
different from the default cm font.

However, I noticed that the agg backend does not account the latex
preamble while caching the tex output, so changing preamble after a
text is cached (after some figure is drawn) does not work.

Regards,

-JJ

2010/4/14 Jon Moore <jon...@ya...>:
> Hi,
>
> Find attched log.txt generated when trying to run your script.  Any
> thoughts?

Sounds like a typo to me.  Please give the new version of crashtest.py
a try, it prints stack tracebacks for each import statement into the
file, so we *should* be able to track it down with this.

The log file created with "#import calendar" commented out is already
20kB large, so maybe too large for the list.  Your will be expected to
be some megabytes I guess :-( but only the last traceback should
matter.

Also I noticed that our import history is *quite* different.  Make a
vimdiff or similar of my old log.txt and the log.txt you provided.
Note that I cannot make a working import because I have no mpl
currently on this MacBook.

Here comes an example output:

import operator, traceback follows:
  File "crashtest.py", line 24, in <module>
    import calendar
  File "crashtest.py", line 18, in new_import
    return old_import(name, *args, **kwargs)
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/calendar.py",
line 10, in <module>
    import locale as _locale
  File "crashtest.py", line 18, in new_import
    return old_import(name, *args, **kwargs)
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/locale.py",
line 202, in <module>
    import re, operator
  File "crashtest.py", line 17, in new_import
    traceback.print_stack()

so far,
hth,
Friedrich

Hi guys and girls :D

I have a problem with my ubuntu server that it is 8.4 version, and that
version only bring matplotlib package 0.6.0 version.

so I have to modify my code to this version

but one thing I can't: the xscale('log') - wich make the graphics in
logaritimic scale of base 10.

how I do that on matplotlib 0.6.0?

or can anyone give me a link for this reference?

thanks in advanced

see ya


Samuel

On Wed, Apr 14, 2010 at 9:35 AM, Ben Axelrod <BAx...@co...> wrote:
> This example shows how to use 2d plots in a 3d plot:
> http://matplotlib.sourceforge.net/examples/mplot3d/2dcollections3d_demo.html
>
> These examples may also help:
> http://matplotlib.sourceforge.net/trunk-docs/examples/mplot3d/contour3d_demo3.html
> http://matplotlib.sourceforge.net/trunk-docs/examples/mplot3d/pathpatch3d_demo.html
>
> -Ben

Yep those are what I needed.  Thanks!

Jeremy

This example shows how to use 2d plots in a 3d plot:
http://matplotlib.sourceforge.net/examples/mplot3d/2dcollections3d_demo.html

These examples may also help:
http://matplotlib.sourceforge.net/trunk-docs/examples/mplot3d/contour3d_demo3.html
http://matplotlib.sourceforge.net/trunk-docs/examples/mplot3d/pathpatch3d_demo.html

-Ben


-----Original Message-----
From: Jeremy Conlin [mailto:jlc...@gm...] 
Sent: Wednesday, April 14, 2010 11:14 AM
To: mat...@li...
Subject: [Matplotlib-users] Can I make a mplot3d PolyCollection Plot with projection on back wall

I want to make a plot similar to this demo:

http://matplotlib.sourceforge.net/examples/mplot3d/polys3d_demo.html

but also make simple line plots on the "back wall" of the plot,
perhaps with the pyplot.plot command.

How can I do this?

Thanks,
Jeremy

------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Matplotlib-users mailing list
Mat...@li...
https://lists.sourceforge.net/lists/listinfo/matplotlib-users

I want to make a plot similar to this demo:

http://matplotlib.sourceforge.net/examples/mplot3d/polys3d_demo.html

but also make simple line plots on the "back wall" of the plot,
perhaps with the pyplot.plot command.

How can I do this?

Thanks,
Jeremy

Unfortunately, the current version of matplotlib requires Python 2.4 or 
later.  You may be able to grab an older version from the archives on 
SourceForge, however, presuming you don't need any of matplotlib's 
recent features or bugfixes.

Mike

On 04/14/2010 07:46 AM, MariaLuisa Caprera wrote:
> Hi,
> I should add some matplotlib functions to my older project developed  
> with python 2.3 for windows.
> I can't find a link to download this extension for my version of Python .
> Please could someone suggest a possible  solution?
> Thanks
>
>
> ------------------------------------------------------------------------------
> Download Intel® Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>    

Hi,

Find attched log.txt generated when trying to run your script.  Any 
thoughts?

Regards

Jon

Friedrich Romstedt wrote:
> 2010/3/12 Jon Moore <jon...@ya...>:
>   
>> I tried getting a colleague of mine to carryout your email but without any
>> joy.  Trying the import commands you suggested my colleague reported .mpl
>> .dates and .pyplot all crashed with the same exception but .mlab worked ok.
>>  The
>> import matplotlib.pylab with the custom import command didn't work as the
>> exception occurs and the console (spyder) shuts down, so cannot see what
>> happened. IPython also does the same, I'd like to try IDLE at some point.
>>
>> My colleague couldn't get the log file stuff to work (maybe as she doesn't
>> normally use python), so I may have to wait until I'm next back there in 4-5
>> weeks to report back properly.
>>     
>
> Ok, maybe send your colleague the script attached.  Simply run:
>
> $ python crashtest.py
>
> It creates ./log.txt with the log output of the packages attempted to
> load.  The last line is the offending import.  I also attach the
> result of my fully successful run.
>
> And when you find something out, maybe send it also to the list.
>
> Lots of success!
> Friedrich
>
>
>
> __________ Information from ESET NOD32 Antivirus, version of virus signature database 5027 (20100414) __________
>
> The message was checked by ESET NOD32 Antivirus.
>
> http://www.eset.com
>
>   



__________ Information from ESET NOD32 Antivirus, version of virus signature database 5027 (20100414) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com

On Wed, Apr 14, 2010 at 13:56, David Kremer <dav...@gm...> wrote:
> Hello, I would like to ask you for an IRC channel in addition to the mailing
> list. If anyone would like to keep it up, it could be great to have a 24h/24h
> support on the irc channel.

FYI, mpl questions are accepted quite well on #scipy on freenode network.

Regards,
-- 
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi

Hello, I would like to ask you for an IRC channel in addition to the mailing 
list. If anyone would like to keep it up, it could be great to have a 24h/24h 
support on the irc channel.

Thank you very much.

David Kremer

   
Hello,

I would like to ask you a way to fill certain regions with a specific motif
when running imshow(). I take the following example :

For certain value, instead of writing directly a numerical result, I would
like to fill my table with 'x'. For other specific values, with 'o'. How take 
care of this with imshow, and how assign a specific color to such a specific 
region ?

I saw something about the spy method. Do you think it's possible to use it
combined with the imshow() capabilities ?

I'm wondering also about the possibility to make logical 'or' on the data, in 
the order to select an area to fill with a custom color.

Thank you very much.

greetings,

David Kremer

Hi,

I should add some matplotlib functions to my older project developed  with python 2.3 for windows.
I can't find a link to download this extension for my version of Python . 

Please could someone suggest a possible  solution?

Thanks

Hi,

Have you solved your problem ? I'm searching matplotlib for python 2.3 under
windows . Can you help me?
Thanks.


Schnappauf, Andreas wrote:
> 
> Hi there,
> I was searching for an older version of matplotlib for using it with
> python 2.2.1 (parts of the project can only be interpreted with this old
> version :().
> Is there a package for an installation under windows (just like the
> current versions)?
> 
> I tried to build 0.80 and 0.87 from the sources and had no success.
> Thx for any help!
> 
> Greetings
> Andreas
> 
> 
> ________________________________
> iSyst Intelligente Systeme GmbH
> Nordostpark 91 | 90411 Nuernberg
> Geschaeftsfuehrer: Prof. Dr.-Ing. Hans Rauch, Christine Rauch, Daniel
> Heinrich
> Sitz der Gesellschaft: Nuernberg
> Registergericht: Amtsgericht Nuernberg HRB 17887
> Steuernr. 241/129/40894 | USt-IdNr.: DE212895677
> 
> ------------------------------------------------------------------------------
> Download Intel® Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
> 

-- 
View this message in context: http://old.nabble.com/Older-Version-of-Matplotlib-for-Python-2.2.1-tp27781592p28231225.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

Hello,

Could someone confirm me if there is any malfunctioning using these simple
figure functions?

plt.figure(figsize=(2,3))

plt.figure(figsize=(5,6))

plt.figure(figsize=(9,15))

plt.figure(figsize=(19,5))

For some reason I can't get Qt4Agg creating last two figures in specified
sizes. (WXAgg works fine.)

matplotlib.__version__
'1.0.svn'

matplotlib.__revision__
'$Revision: 8226 $'

from PyQt4 import QtCore
QtCore.PYQT_VERSION_STR
'4.7'

Thanks

-- 
Gökhan

>
> Is it fairly easy to put something like this together using all the
> offsetbox tools and fancy arrows?

I tried to cook up something similar to what you described. See the
attached file.
Well, I would not say it is fairly easy, but not that difficult either I hope.
The demo requires svn r8227 to work correctly.

I think the demo is rather self-explanatory. But let me know if there
is anything that are not clear.
Regards,

-JJ


>
> Thanks,
> JDH
>
> ------------------------------------------------------------------------------
> Download Intel® Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

On 2010-04-13 16:55 PM, K.-Michael Aye wrote:

> Trying Unison via the GMane NNTP now, but weird that nabble has your
> last answer already for long time, whereas GMane still does not show
> it. Does the NNTP pull the mailing lists on a low frequency.

The latency is variable, but it's been getting pretty bad recently.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
  that is made terrible by our own mad attempt to interpret it as though it had
  an underlying truth."
   -- Umberto Eco

On 2010-04-13 18:13:40 +0200, K. -Michael Aye said:

>> 
>> 
>> On 2010-04-13 10:18 AM, K. -Michael Aye wrote:
>>> Dear all,
>>> 
>>> maybe this should go to the Enthought list, but as the failure is 
>>> directly related to the pylab switch of ipython, I thought I try it 
>>> here first:
>>> 
>>> On OSX I have trouble with using the pylab switch for ipython after I 
>>> copied the gdal.pth into the Enthought site-packages folder (to be able 
>>> to use my KyngChaos GDAL Frameworks inside the Enthought Python).
>>> 
>>> The gdal.pth does the following to the sys.path:
>>> import sys; 
>>> sys.path.insert(0,'/Library/Frameworks/GDAL.framework/Versions/1.7/Python/site-packages')

and 
>>> 
>>> in that folder there is:
>>> 
>>> -rw-rw-r--   1 root  admin   128B  8 Feb 20:52 gdal.py
>>> -rw-r--r--   1 root  admin   274B  3 Mar 23:20 gdal.pyc
>>> -rw-rw-r--   1 root  admin   143B  8 Feb 20:52 gdalconst.py
>>> -rw-r--r--   1 root  admin   304B  3 Mar 23:20 gdalconst.pyc
>>> -rw-rw-r--   1 root  admin   147B  8 Feb 20:52 gdalnumeric.py
>>> -rw-r--r--   1 root  admin   309B  3 Mar 23:20 gdalnumeric.pyc
>>> drwxrwxr-x  42 root  admin   1.4K  3 Mar 23:20 numpy
>>> -rw-rw-r--   1 root  admin   125B  8 Feb 20:52 ogr.py
>>> -rw-r--r--   1 root  admin   286B  3 Mar 23:20 ogr.pyc
>>> drwxrwxr-x  21 root  admin   714B  3 Mar 23:20 osgeo
>>> -rw-rw-r--   1 root  admin   125B  8 Feb 20:52 osr.py
>>> -rw-r--r--   1 root  admin   286B  3 Mar 23:20 osr.pyc
>>> 
>>> Maybe the double import of a potentially different numpy compared to 
>>> the Enthought numpy creates the Bus Error?
>> 
>> Not so much a double import. Only one version ever gets imported, but the GDAL
>> Python bindings expect its version and matplotlib expects another version.
>> 
>>> If so, how can I avoid it?
>> 
>> You would have to rebuild the GDAL Python bindings against Enthought's numpy.
>> 
> But why does everything work fine, when I start an Enthought ipython 
> withOUT the -pylab switch?
> Importing 'from osgeo import gdal' and using it works fine in this case 
> (Tried ReadAsArray from a gdal dataset and imshow'ed it without 
> problems, apart from that I had to call show() because of the lack of 
> the -pylab switch, but other than that, fine).

Sorry, seems that I had messed up my config somehow, now GDAL does not 
work anymore inside the EPD Python. Hmm, have to make some clean 
tests...

> 
> PS.: Sorry for the mail-list noob question, but how can I nicely reply 
> to your answer like you replied to my question, with 'Robert Kern 
> wrote' and so on? There's no reply possible on sourceforge and the 
> digest contains obviously many emails, so how do you do this? ;)

Trying Unison via the GMane NNTP now, but weird that nabble has your 
last answer already for long time, whereas GMane still does not show 
it. Does the NNTP pull the mailing lists on a low frequency.
Man, I think it's 10 years ago or so since I have used NNTP. Used it a 
lot in my first net years, 94/95 ;)
Thanks for the GMane tip, the nabble reply web interface does not even 
quote properly, very strange thing, or i do something wrong.

Have you been meaning to prepare an abstract to submit for a SciPy 2010
specialized track (http://conference.scipy.org/scipy2010/papers.html#tracks 
)?
Didn't find the time? Well you're in luck.
This weekend, we had technical issues with the email submissions for the
specialized tracks. In light of the inconvenience, we've decided to  
extend
the deadline an additional two weeks until Sunday, April 25th.

If you have an abstract ready for one of the four specialized tracks,  
please
use the links below to submit it to the program chair. If you previously
submitted one and didn't receive confirmation that we received it, it  
would
be a great idea to submit it again to ensure we get it.

   * Biomedical/bioinformatics chaired by Glen Otero, Dell
     submit/contact: 201...@sc...

   * Financial analysis chaired by Wes McKinney, AQR Capital Management
     submit/contact: 201...@sc...

   * Geophysics chaired by Alan Jackson, Shell
     submit/contact: 201...@sc...

   * Parallel processing & cloud computing co-chaired by Ken Elkabany,
     PiCloud & Brian Granger, CalPoly
     submit/contact: 201...@sc...


Main Conference Submissions
Submissions for the main SciPy 2010 conference closed Sunday. Thanks to
everyone who submitted. We'll announce the accepted talks Tuesday April
20th.

Student Sponsorships
If you're an academic and contribute to SciPy or related projects,  
make sure
to apply for one of our student sponsorships. The deadline to apply is
April 18th. We are also accepting nominations.
http://conference.scipy.org/scipy2010/student.html

Don't forget to register...
Registrations are coming in pretty steadily now. Remember that to get  
early
registration prices you need to [8]register before May 10th!
https://conference.scipy.org/scipy2010/registration.html

The SciPy 2010 Team
@SciPy2010 on Twitter

--
Amenity Applewhite
Enthought, Inc.
Scientific Computing Solutions
www.enthought.com

Drain, Theodore R (343P) wrote:
> On Linux:
> 
> import pylab as p
> p.plot( [1,2,3] )
> p.savefig( 'test.ps' )
> p.savefig( 'test.png' )
> 
> ls -l test.*
> -rw-r--r--  1 --- ----- 17236 Apr 13 10:32 test.png
> -rw-------  1 --- -----  8640 Apr 13 10:32 test.ps
> 

> This is with a umask of 022. It looks to me like this is caused by
> the
> ps backend creating a temp file and then moving it to the requested file
> name. The permissions in the /tmp directory are generally different than
> a normal file system.
> 
> Any thoughts on whether this is bug or not? It seems to me that using
> a temp file is an implementation detail and calling savefig should
> always create files w/ the user's permission settings.


Yes, I think it is a bug.

> I'm not sure how to "fix" this - the normal copy command preserves
> the permissions.  I guess the temp file could be opened and the data
> read and then copied to the requested file.
> 

An alternative, avoiding the copy operation, is to replace

shutil.move(tmpfile, outfile)

with

open(outfile, 'w')
mode = os.stat(outfile).st_mode
shutil.move(tmpfile, outfile)
os.chmod(outfile, mode)

Eric

> Ted
> ------------------------------------------------------------------------------
> Download Intel® Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On Linux:

import pylab as p
p.plot( [1,2,3] )
p.savefig( 'test.ps' )
p.savefig( 'test.png' )

ls -l test.*
-rw-r--r--  1 --- ----- 17236 Apr 13 10:32 test.png
-rw-------  1 --- -----  8640 Apr 13 10:32 test.ps

This is with a umask of 022.   It looks to me like this is caused by the ps backend creating a temp file and then moving it to the requested file name.  The permissions in the /tmp directory are generally different than a normal file system.  

Any thoughts on whether this is bug or not?  It seems to me that using a temp file is an implementation detail and calling savefig should always create files w/ the user's permission settings.

I'm not sure how to "fix" this - the normal copy command preserves the permissions.  I guess the temp file could be opened and the data read and then copied to the requested file.  

Ted

On 2010-04-13 11:13 AM, K. -Michael Aye wrote:
>>
>> On 2010-04-13 10:18 AM, K. -Michael Aye wrote:
>>> Dear all,
>>>
>>> maybe this should go to the Enthought list, but as the failure is directly related to the pylab switch of ipython, I thought I try it here first:
>>>
>>> On OSX I have trouble with using the pylab switch for ipython after I copied the gdal.pth into the Enthought site-packages folder (to be able to use my KyngChaos GDAL Frameworks inside the Enthought Python).
>>>
>>> The gdal.pth does the following to the sys.path:
>>> import sys; sys.path.insert(0,'/Library/Frameworks/GDAL.framework/Versions/1.7/Python/site-packages')
>>>
>>> and in that folder there is:
>>>
>>> -rw-rw-r--   1 root  admin   128B  8 Feb 20:52 gdal.py
>>> -rw-r--r--   1 root  admin   274B  3 Mar 23:20 gdal.pyc
>>> -rw-rw-r--   1 root  admin   143B  8 Feb 20:52 gdalconst.py
>>> -rw-r--r--   1 root  admin   304B  3 Mar 23:20 gdalconst.pyc
>>> -rw-rw-r--   1 root  admin   147B  8 Feb 20:52 gdalnumeric.py
>>> -rw-r--r--   1 root  admin   309B  3 Mar 23:20 gdalnumeric.pyc
>>> drwxrwxr-x  42 root  admin   1.4K  3 Mar 23:20 numpy
>>> -rw-rw-r--   1 root  admin   125B  8 Feb 20:52 ogr.py
>>> -rw-r--r--   1 root  admin   286B  3 Mar 23:20 ogr.pyc
>>> drwxrwxr-x  21 root  admin   714B  3 Mar 23:20 osgeo
>>> -rw-rw-r--   1 root  admin   125B  8 Feb 20:52 osr.py
>>> -rw-r--r--   1 root  admin   286B  3 Mar 23:20 osr.pyc
>>>
>>> Maybe the double import of a potentially different numpy compared to the Enthought numpy creates the Bus Error?
>>
>> Not so much a double import. Only one version ever gets imported, but the GDAL
>> Python bindings expect its version and matplotlib expects another version.
>>
>>> If so, how can I avoid it?
>>
>> You would have to rebuild the GDAL Python bindings against Enthought's numpy.
>>
> But why does everything work fine, when I start an Enthought ipython withOUT the -pylab switch?
> Importing 'from osgeo import gdal' and using it works fine in this case (Tried ReadAsArray from a gdal dataset and imshow'ed it without problems, apart from that I had to call show() because of the lack of the -pylab switch, but other than that, fine).

Hmm, don't know. Getting a gdb traceback for the bus error would help identify 
the problem.

> PS.: Sorry for the mail-list noob question, but how can I nicely reply to your answer like you replied to my question, with 'Robert Kern wrote' and so on? There's no reply possible on sourceforge and the digest contains obviously many emails, so how do you do this? ;)

I use an NNTP newsreader to read this list via GMane, but you can just change 
your subscription to not use the digest. Scroll down to the bottom of this page 
to log in and edit your delivery options:

https://lists.sourceforge.net/lists/listinfo/matplotlib-users

You will get every message in your inbox individually. You should do this if you 
are going to be replying to messages. Please consider the digest as read-only.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
  that is made terrible by our own mad attempt to interpret it as though it had
  an underlying truth."
   -- Umberto Eco