matplotlib-users Mailing List for matplotlib

On Tuesday 26 July 2005 01:11 pm, Mark Bakker wrote:
> I have the same experience that eps files seem to be much larger than
> you expect, or than produced by other programs for similar plots.
> Is it maybe because fonts are embedded in the eps file?
> Anybody know?

It is because the fonts are embedded in the eps file.

-- 

Darren

Hi John,
    Thanks again for the quick responce. Sorry about not doing the 
simple test earlier. I installed the latest version this time
and used the script you suggested (.py file is attached). Here's the output:

C:\proj\compclust\win32\compclustshell>C:\Python23\python.exe 
matplotlibtest.py
--verbose-helpful
matplotlib data path C:\Python23\share\matplotlib
$HOME=C:\Documents and Settings\King
CONFIGDIR=C:\Documents and Settings\King\.matplotlib
loaded rc file C:\Python23\share\matplotlib\matplotlibrc
matplotlib version 0.83.1
verbose.level helpful
interactive is False
platform is win32
numerix Numeric 23.8
font search path ['C:\\Python23\\share\\matplotlib']
Assertion failed: ob_refcnt == 0, file CXX\cxx_extensions.cxx, line 1031

This application has requested the Runtime to terminate it in an unusual 
way.
Please contact the application's support team for more information.

-Brandon King

P.S. Thank you for the information about the documented API changes! =o)

John Hunter wrote:

>>>>>>"Brandon" == Brandon King <ki...@ca...> writes:
>>>>>>            
>>>>>>
>
>    Brandon> Hi John, Thanks for the quick reply. In this case we are
>    Brandon> using windows, to make a windows build of our code.  I
>    Brandon> removed site-packages/matplotlib and re-installed it. I
>    Brandon> had to install version 0.80, because an API change in
>    Brandon> either .81 or .82 broke some of our code and we haven't
>    Brandon> had a chance to fix that.  I ran our script as you
>    Brandon> said... here's the output with the same error:
>
>You can still install 0.83 and test with that (later you can revert to
>the older version if need be).  This will provide helpful diagnostic
>information.  So once again, please remove site-packages/matplotlib
>and install the latest and test with a minimal script, eg
>
>    import pylab
>    pylab.plot([1,2,3])
>    pylab.show()
>
>with the verbose-helpful flags.
>
>Note also you can see http://matplotlib.sf.net/API_CHANGES for the
>migration path from 0.80 to 0.83.  It's likely as simple as a search
>and replace from set to setp, or something like that.  The API is
>fairly stable and all changes are documented there.
>
>JDH
>
>
>  
>

>>>>> "Brandon" == Brandon King <ki...@ca...> writes:

    Brandon> Hi John, Thanks for the quick reply. In this case we are
    Brandon> using windows, to make a windows build of our code.  I
    Brandon> removed site-packages/matplotlib and re-installed it. I
    Brandon> had to install version 0.80, because an API change in
    Brandon> either .81 or .82 broke some of our code and we haven't
    Brandon> had a chance to fix that.  I ran our script as you
    Brandon> said... here's the output with the same error:

You can still install 0.83 and test with that (later you can revert to
the older version if need be).  This will provide helpful diagnostic
information.  So once again, please remove site-packages/matplotlib
and install the latest and test with a minimal script, eg

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

with the verbose-helpful flags.

Note also you can see http://matplotlib.sf.net/API_CHANGES for the
migration path from 0.80 to 0.83.  It's likely as simple as a search
and replace from set to setp, or something like that.  The API is
fairly stable and all changes are documented there.

JDH

Hi John,
    Thanks for the quick reply. In this case we are using windows, to 
make a windows build of our code.
I removed site-packages/matplotlib and re-installed it. I had to install 
version 0.80, because an API change
in either .81 or .82 broke some of our code and we haven't had a chance 
to fix that.
    I ran our script as you said... here's the output with the same error:

C:\>C:\Python23\python.exe compclustweb-local.py --verbose-helpful
matplotlib data path C:\Python23\share\matplotlib
loaded rc file C:\Python23\share\matplotlib\.matplotlibrc
matplotlib version 0.80
verbose.level helpful
interactive is False
platform is win32
numerix Numeric 23.8
font search path ['C:\\Python23\\share\\matplotlib']
Assertion failed: ob_refcnt == 0, file CXX\cxx_extensions.cxx, line 1031

This application has requested the Runtime to terminate it in an unusual 
way.
Please contact the application's support team for more information.

Most of our code (all python code) is open source, so if you want to 
look at it you can.
The only problem is that do to the powers that be, some of c code is 
only available in binary forms
(were working on open sourcing that as well). Let me know if your 
interested.

Thank you for your help!

-Brandon King

John Hunter wrote:

>>>>>>"Brandon" == Brandon King <ki...@ca...> writes:
>>>>>>            
>>>>>>
>
>    Brandon> Hi All, I ran into the following error:
>
>    Brandon> Assertion failed: ob_refcnt == 0, file
>    Brandon> CXX\cxx_extensions.cxx, line 1031
>
>I assume you are using win32.
>
>Please remove site-packages/matplotlib and reinstall the latest
>version.  Then run your test script with
>
>  > python myscript.py --verbose-helpful
>
>and if the error persists, post any output.
>
>Thanks,
>JDH
>
>
>-------------------------------------------------------
>SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
>from IBM. Find simple to follow Roadmaps, straightforward articles,
>informative Webcasts and more! Get everything you need to get up to
>speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
>_______________________________________________
>Matplotlib-users mailing list
>Mat...@li...
>https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>  
>

>>>>> "Brandon" == Brandon King <ki...@ca...> writes:

    Brandon> Hi All, I ran into the following error:

    Brandon> Assertion failed: ob_refcnt == 0, file
    Brandon> CXX\cxx_extensions.cxx, line 1031

I assume you are using win32.

Please remove site-packages/matplotlib and reinstall the latest
version.  Then run your test script with

  > python myscript.py --verbose-helpful

and if the error persists, post any output.

Thanks,
JDH

Hi All,
  I ran into the following error:

Assertion failed: ob_refcnt == 0, file CXX\cxx_extensions.cxx, line 1031

I searched google and found the e-mail post from 2004-06-23 which I've included
at the bottom of this e-mail. I was wondering if anyone has figured out the 
problem? I'm guessing this is still likely to be a bug related to matplotlib, 
but I don't know for sure.

What I do know is that why I try to import matplotlib directly into python,
I don't get the error (see code below).

 > >>> import matplotlib
 > >>> matplotlib.use("TkAgg")
 > >>> from matplotlib.pylab import *

But when we run our program which uses matplotlib, we get that error. I'll try to debug
our program and figure out where it seems to be crashing. If it's in matplotlib code, 
I'll dig in deeper to try to figure out where. I'll post again when I have more information.

I mainly wanted to give everyone a heads up and also see if anyone can give me some pointers.

-Brandon King



On Wed, 2004-06-23 at 12:42, Gregory Lielens wrote:
 > Hello All,
 > I just tried to install matplotlib on a WinXP computer, after a
 > successfull install on Linux.
 > The problem I encounter is the following (transcript of an
 > interractive session): 
 >  
 > ---
 > C:\Python23>python
 > Python 2.3.4 (#53, May 25 2004, 21:17:02) [MSC v.1200 32 bit (Intel)]
 > on win32
 > Type "help", "copyright", "credits" or "license" for more information.
 > >>> import matplotlib
 > >>> matplotlib.use("TkAgg")
 > >>> from matplotlib.matlab import *
 > Assertion failed: ob_refcnt == 0, file CXX/cxx_extensions.cxx, line
 > 1031
 >  
 > This application has requested the Runtime to terminate it in an
 > unusual way.
 > Please contact the application"s support team for more information.
 >  
 > C:\Python23>
 >  
 > ---
 >  
 > This work on Linux, and I can not find any hint in the doc to a
 > backend/python interpreter incompatibility explaining this error.
 >  
 > I use  numarray version 0.9 (win installer), python 2.3.4 (as shown in
 > the session transcript), and matplotlib 0.54.2, last windows installer
 > (the 2004/06/08 21:54:42 build which I think correct some problems
 > with numarray)
 >  
 > Now I am wondering if I do something I shouldn"t, if my system or
 > installation is responsible, or if it is a bug in this particular
 > matplotlib distribution...Any idea? 

I have the same experience that eps files seem to be much larger than
you expect, or than produced by other programs for similar plots.
Is it maybe because fonts are embedded in the eps file?
Anybody know?

Thanks, Mark

>>>>> "Christopher" == Christopher Hart <chr...@gm...> writes:


    Christopher> Effectively, I have a situation where I have multiple
    Christopher> "traces" that I would like to display. Each trace
    Christopher> requires a separate subplot with a different Y-axis,
    Christopher> but all subplots should share the exact same X-axis.
    Christopher> I'd like to be able to interactively explore these
    Christopher> plots without fear of the subplots becoming
    Christopher> out-of-sync.

Use the sharex or sharey attribute of the Axes

    ax1 = subplot(211)
    ax1.plot(something)

    ax2 = subplot(212, sharex=ax1)
    ax2.plot(something, else)

See examples/shared_axis_demo.py for a complete example.  Now when you
pan or zoom on either subplot, the two x axes remain in sync.

Neat stuff! Thanks to Baptiste Carvello for providing this feature.

JDH

Hi,

I was wondering if there is an "easy" way of linking the x-axis (or
y-axis for that matter) of two subplots.  Using the "classic" toolbar
you can select to have pan and zoom events effect all or a specific
subset of axis.  Can you do the same using the "toolbar2" mode? Is
there a function to also provide this functionality programmatically,
so if I zoom or pan one subplot, the other subplots follow suite.

Effectively, I have a situation where I have multiple "traces" that I
would like to display. Each trace requires a separate subplot with a
different Y-axis, but all subplots should share the exact same X-axis.
  I'd like to be able to interactively explore these plots without
fear of the subplots becoming out-of-sync.

Thanks,

Chris

N. Volbers wrote:
> 
>> Hmmm now that you mention antialiasing ... I turned it off in my 
>> .matplotlibrc but my plots (GTKAgg on Linux) are still antialiased. I 
>> had the same problem on Win (TkAgg) but didn't bother too much. I 
>> remember that there was a discussion before about this. I'm using mpl 
>> 0.81. Would upgrading fix this issue?
> 
> 
> 
> I tested the python script on my mpl, version 0.83.1 with no custom 
> matplotlibrc. To my surprise, I got the result you were talking about.  
> Then I opened the resulting eps with gv on linux.  There is an option 
> "antialias", which when you turn it on, will cause this color mixing.
> 
> To demonstrate this, I wrote a little gnuplot script that produces a 
> similar output.  If you look at the attached file with gv and toggle 
> antialiasing, you will notice the same effect as with mpl.

Yes I can varify that (on Win) with gsview (where turning off AA is done 
by setting Media > Display Settings > 'Text Alpha' and 'Graphics Alpha' 
to 1 bit). OK so it really seems to be no mpl issue. Just like Jhon 
mentioned :)

But one problem remains: I'm not able to turn off AA in .matplotlibrc
If anyone could point me in the right direction (upgrading, other 
backend?) ....

> 
> On a side note, when comparing these two graphs I noticed that the 
> gnuplot generated eps is only 15 K while the mpl generated eps has a 
> size of 300 K.  Why is this so?
> 

I guess the dpi setting. But playing around with different dpi settings

	figure(dpi=<number>); plot(...); savefig(...)

or

	plot(...); savefig(...,dpi=<number>)

_always_ produces a .eps of 365 kb.

> Regards,
> 
> Niklas.
> 

cheers,
steve

-- 
Women are like cell phones. They like to be held and talked to, but push 
the wrong button, and you'll be disconnected.

> Hmmm now that you mention antialiasing ... I turned it off in my 
> .matplotlibrc but my plots (GTKAgg on Linux) are still antialiased. I 
> had the same problem on Win (TkAgg) but didn't bother too much. I 
> remember that there was a discussion before about this. I'm using mpl 
> 0.81. Would upgrading fix this issue?


I tested the python script on my mpl, version 0.83.1 with no custom 
matplotlibrc. To my surprise, I got the result you were talking about.  
Then I opened the resulting eps with gv on linux.  There is an option 
"antialias", which when you turn it on, will cause this color mixing.

To demonstrate this, I wrote a little gnuplot script that produces a 
similar output.  If you look at the attached file with gv and toggle 
antialiasing, you will notice the same effect as with mpl.

On a side note, when comparing these two graphs I noticed that the 
gnuplot generated eps is only 15 K while the mpl generated eps has a 
size of 300 K.  Why is this so?

Regards,

Niklas.

>>>>> "Malte" == Malte Marquarding <Mal...@cs...> writes:

    Malte> Ditto here.  Seems to have been introduced somewhere from
    Malte> version > 0.80.

    Malte> version 0.8 works for me. version 0.82 doesn't.

This was a bug introduce in 0.82, and it was fixed in 0.83.

Thanks,
JDH

Ditto here.
Seems to have been introduced somewhere from version > 0.80.

version 0.8 works for me. version 0.82 doesn't.

Cheers,
Malte.