matplotlib-users Mailing List for matplotlib

On 2007年12月17日, Michael Droettboom apparently wrote:
> You can generate a PDF file with matplotlib, and then run 
> it through "acroread -toPostScript foo.pdf", and get 
> a reasonably well-optimized Postscript file. 
I did not know that.
Thanks!
Alan

Well, I hate to provide the disappointing answer, but it really doesn't 
work for Postscript. (The name pstricks has perhaps become misleading 
over time...)
It inserts special codes in the Postscript that aren't part of the Ps 
standard, but that ps2pdf14 is able to convert into the correct Pdf 
commands to handle alpha-blending. I don't see any advantage to that 
kind of workflow (with matplotlib) over just going directly to PDF.
There are other tricks to "fake" transparency in Postscript, for 
example, by calculating the polygon intersections, etc., but that is 
significant work, and hard to do in a general way (i.e. anything 
overlapping anything), without writing a full-fledged geometry framework.
Fortunately, Adobe Reader already has such a thing. You can generate a 
PDF file with matplotlib, and then run it through "acroread 
-toPostScript foo.pdf", and get a reasonably well-optimized Postscript 
file. Unfortunately, ghostscript 7.07 doesn't seem to do this -- it can 
convert a pdf to a ps file, but the alphablended stuff appears as a 
raster image. Maybe newer versions (or poppler etc.) are able to do 
this, but I don't have them on my machine. It would be nice to know of 
an open source alternative, but Adobe Reader is at least free-as-in-beer.
Cheers,
Mike
Alan G Isaac wrote:
> On 2007年12月17日, Tom Johnson apparently wrote:
>> pstricks (tex) provides transparency... 
>> http://tug.org/PSTricks/main.cgi?file=Examples/Colors/colors#transparency 
>> http://tug.org/PSTricks/main.cgi?file=pst-plot/3D/examples#coor 
> 
> 
> Impressive. How is it done?
> 
> Cheers,
> Alan Isaac
> 
> 
> 
> 
> -------------------------------------------------------------------------
> SF.Net email is sponsored by:
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services
> for just about anything Open Source.
> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
-- 
Michael Droettboom
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA

Thanks for the reply. I fixed my problem and thought I would post my 
findings. Basically, I never did find out what the problem was with 
those particular version was, but upgrading the packages eventually 
fixed it. Of course, that meant compiling numpy, python, pango, 
cairo.... etc. Took forever, but it did eventually work.
Thanks for all of your advice!
JK
John Hunter wrote:
> On Dec 5, 2007 12:53 PM, Jonathan King <ki...@mi...> wrote:
> 
>> i am having trouble installing matplotlib from source and thought i
>> would post my issue here.
>> i download matplotlib version 0.91.0
>> 
>
> It looks possibly like a numpy problem (you have a pretty old version installed)
>
> 
>> /usr/lib/python2.3/site-packages/numpy/core/include/numpy/ufuncobject.h:9:
>> error: 'intp' has not been declared
>> 
>
> Consider first upgrading to the latest numpy.
>
> JDH
> 

In the interest of fixing this bug for others, I wonder if you wouldn't 
mind testing something for me (since I don't have Windows handy, and 
even if I did, it wouldn't be *your* Windows environment...)
The only different between the provided matplotlibrc and the built-in 
defaults seems to be font.weight. In the matplotlibrc it is "medium", 
and in the defaults it is "normal". If you uncomment only 
"font.weight", and leave the others commented out, do things work for you?
Cheers,
Mike
Orest Kozyar wrote:
>> Not sure if this is your issue but I had problems with fonts until I
>> uncommented the RC file in the fonts section of it. Matplotlib ships with
>> the RC commented out. The file matplotlibrc is found in the
>> matplotlib\mpl-data folder. You can just edit it with Notepad and uncomment
>> the lines regarding fonts and try it again.
> 
> That worked perfectly. Thanks!
> 
> -------------------------------------------------------------------------
> SF.Net email is sponsored by:
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services
> for just about anything Open Source.
> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
-- 
Michael Droettboom
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA

On 2007年12月17日, Tom Johnson apparently wrote:
> pstricks (tex) provides transparency... 
> http://tug.org/PSTricks/main.cgi?file=Examples/Colors/colors#transparency 
> http://tug.org/PSTricks/main.cgi?file=pst-plot/3D/examples#coor 
Impressive. How is it done?
Cheers,
Alan Isaac

Hi all,
I wasn't sure if this should be sent to matplotlib or ipython. I'm
got a number of questions, and any help would be greatly appreciated.
The docstring for draw_if_interactive says:
"""This should be overriden in a windowing environment if drawing
should be done in interactive python mode"""
(there should be two d's in overridden)
First, I'm not quite sure what this means....doesn't 'interactive'
imply 'interactive python mode'? So shouldn't this function always
need overriding (strange that this doesn't have two d's)?
The definition is:
 if matplotlib.is_interactive():
 figManager = Gcf.get_active()
 if figManager is not None:
 figManager.canvas.draw()
Now, when I load with ipython -pylab, the definition is:
 def wrapper(*args,**kw):
 wrapper.called = False
 out = func(*args,**kw)
 wrapper.called = True
 return out
Ipython says the definition is in genutils.py (ipython) but I can't
find it in there, and I don't know what "func" is. This is the
ipython part of my email: What does this function mean/do...and where
can I find where 'func' is defined.
Finally, the real reason for my email. I've been writing functions
which perform more complicated plot commands. In general, I don't
want the substeps to be shown, so I call matplotlib.interactive(False)
and then restore the state at the end of the function. Then I call
pylab.draw_if_interactive(). The first issue is that this doesn't
really work. Essentially this is what I am doing:
$ ipython -pylab
In [1]: ioff()
In [2]: plot(range(10))
In [3]: ion()
In [4]: draw_if_interactive()
On doing this, nothing shows up, unless I call show(). But the
following works without calling show....
$ ipython -pylab
In [1]: plot(range(10))
Why?
Assuming I can get this to work...it seems like a good solution so
long as my functions are called from "ipython -pylab", but I would
also like to be able to call such functions inside a GUI....and I am
concerned with the ramifications of the pylab.draw_if_interactive()
call. The only pylab command in the plot functions is the
draw_if_interactive...and my (wxPython) GUI uses OO matplotlib
throughout. Is this an issue? Do I need to overwrite this function?
If so, what needs to change so that I can use my functions in a GUI
(where I manually call draw) and in ipython. The quickest solution is
to have each function accept an keyword which tells it whether or not
to call draw_if_interactive()...but this is pain...and doesn't seem
very elegant. None of the matplotlib functions call
draw_if_interactive()...so perhaps there is another way to temporarily
turn off interactive mode and then restore the state.
Thanks.

On Dec 14, 2007 11:28 AM, Jeff Whitaker <js...@fa...> wrote:
> Mike: Postscript doesn't support alpha transparency. It might work
> with PDF though.
pstricks (tex) provides transparency...
http://tug.org/PSTricks/main.cgi?file=Examples/Colors/colors#transparency
http://tug.org/PSTricks/main.cgi?file=pst-plot/3D/examples#coor
Is there any way to get similar hacks into matplotlib? I'm guessing
this is a lot of work though.

Hi Michael,
Thanks for the reply. A standalone example .py file is attached,
including a CSV file that it loads.
The backend is 'GTKAgg', and numerix is 'numpy'; settings for each of
these are all that is in my ~/.matplotlib/matplotlibrc file.
Cheers
JP
Michael Droettboom wrote:
> Can you provide a standalone example that exhibits this problem? I'm
> not able to reproduce this with 0.90.1 and the imshow examples
> (image_demo.py, image_origin.py).
>
> Which backend are you using?
>
> My only wild guess is -- with the Agg backend (possibly others), the
> image is stored flipped from how Agg wants to render it. So, the
> image is flipped (not the data, just the stride), then it is drawn,
> then it is flipped back. If there were an exception in the middle
> there, the number of flips may not remain even. Are there any
> exceptions being output? (These exceptions, since they're inside a
> GUI callback wouldn't necessarily halt the application). Beyond that,
> you could trace calls to im.flipupd_out and make sure they are even in
> number.
>
> Cheers,
> Mike
>
> John Pye wrote:
>> Hi all
>>
>> I have a problem with 'imshow' under matplotlib 0.90.1-2ubuntu1 on
>> ubuntu 7.10. I have an 'incidence matrix' created using the 'imshow'
>> command, and it works well except for the fact that sometimes when
>> resizing my window, the incidence matrix flips upside-down.
>>
>> Can I control this flipping of the incidence matrix? it seems to me to
>> be something that has just happened in a recent release of matplotlib,
>> as my application didn't used to show this problem.
>>
>> FYI The code that I'm using to drive mpl is here:
>> http://ascendcode.cheme.cmu.edu/viewvc.cgi/code/branches/extfn/pygtk/incidencematrix.py?view=markup
>>
>>
>> Cheers
>> JP
>>
>>
>>
>>
>> -------------------------------------------------------------------------
>>
>> SF.Net email is sponsored by:
>> Check out the new SourceForge.net Marketplace.
>> It's the best place to buy or sell services
>> for just about anything Open Source.
>> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
>>
>> _______________________________________________
>> Matplotlib-users mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>