matplotlib-users Mailing List for matplotlib (Page 13)

On Mon, Aug 9, 2010 at 8:48 AM, Filipe Fernandes <oc...@gm...> wrote:
>
> The docstring for sunplots has the "sharex(y) option", but my "ax" are
> being created now!? Is that a chicken and egg problem? Or I'm failing to
> understand the logic here.

Don't worry: when you indicate that you want to share axes (x or y),
subplots() tracks that and enables it as it creates the axis objects.
Give it a try and see if it works :)

Cheers,

f

On Aug 9, 2010, at 8:17 AM, Friedrich Romstedt wrote:

> 2010/8/7 Russell E. Owen <ro...@uw...>:
>> In article <SNT...@ph...>,
>>  Stephen T. <obs...@ho...>
>>  wrote:
>>
>>> Hi, I am having trouble installing matplotlib. I have OS X 10.5  
>>> with Python
>>> 2.6
>>>  downloaded and installed from python.org.
>>>  (10.5 came with Apple
>>>  Python 2.5). I've also installed NumPy and SciPy for Python 2.6.
>>
>> I encourage you to try the matplotlib 1.0.0 binary installer. It was
>> built to use the python.org Python (versions are available for Python
>> 2.5 and 2.6; the latter did have a known problem with permissions  
>> on the
>> icon files; you can fix that manually or a new installer with that
>> problem fixed should be up now or shortly).
>
> I agree trying that will be an option.
>
>> Another option is to wade through the build instructions. I've got  
>> some
>> here:
>> <http://www.astro.washington.edu/users/rowen/BuildingMatplotlibForMac.html 
>> >
>
> Hmm, as I see it this instructions are for building a binary
> distribution package of matplotlib, I think this is not what Stephen
> wants?

The instructions are indeed geared towards building a binary  
installer, but the final result is a build, plus it is possible to  
weed out the installer-specific elements.

-- Russell

Hello list,

I enjoyed the new feature:
fig, axarr = plt.subplots(2, 2)

However, I failed to understand how to use the shared axis option with 
this new feature.

The docstring for sunplots has the "sharex(y) option", but my "ax" are 
being created now!? Is that a chicken and egg problem? Or I'm failing to 
understand the logic here.

Thanks, Filipe.

2010/8/7 Russell E. Owen <ro...@uw...>:
> In article <SNT...@ph...>,
>  Stephen T. <obs...@ho...>
>  wrote:
>
>> Hi, I am having trouble installing matplotlib. I have OS X 10.5 with Python
>> 2.6
>>  downloaded and installed from python.org.
>>  (10.5 came with Apple
>>  Python 2.5). I've also installed NumPy and SciPy for Python 2.6.
>
> I encourage you to try the matplotlib 1.0.0 binary installer. It was
> built to use the python.org Python (versions are available for Python
> 2.5 and 2.6; the latter did have a known problem with permissions on the
> icon files; you can fix that manually or a new installer with that
> problem fixed should be up now or shortly).

I agree trying that will be an option.

> Another option is to wade through the build instructions. I've got some
> here:
> <http://www.astro.washington.edu/users/rowen/BuildingMatplotlibForMac.html>

Hmm, as I see it this instructions are for building a binary
distribution package of matplotlib, I think this is not what Stephen
wants?

Dear Stephen, did the mpl-1.0.0 thingy work for you, do you have
problems with the icon files, if yes, the instruction inspired by
Russell:

$ chmod -R a+r [wherever they're living]

will fix it I believe.  The -R option can be left alone when there is
only one level of file hierarchy to fix, but I'm not sure what
applies.

I really hope we don't have to go through the /opt stuff, because I'm
really unsure about how to fix this.  Maybe check your
$LD_LIBRARY_PATH or similar (I don't recall precisely), if it contains
the /opt stuff, we're happy, because it's easy to fix.  Say easier.  I
believe it will be a pain at all anyway.  Best would probably be to
start with a new, fresh and empty mpl build directory and to go
through each step with feedback from the list ... because I don't know
precisely what will happen.

Friedrich

P.S.: I've been busy the past few days, please excuse me.

2010/8/9 John Hunter <jd...@gm...>:
> On Sun, Aug 8, 2010 at 8:03 PM, Friedrich Romstedt
> <fri...@gm...> wrote:
>> P.S.: You can also try
>> http://matplotlib.sourceforge.net/examples/pylab_examples/alignment_test.html
>> , I believe there was some other manual around also covering the
>> baseline option, but I cannot find it now.
>
> The "codex" option on the search page helps with this; you can easily
> search for "codex baseline"::
>
>  * the default  search page: http://matplotlib.sourceforge.net/search.html
>
>  * what is codex?:
> http://matplotlib.sourceforge.net/faq/howto_faq.html#search-examples
>
>  * an example using "codex baseline":
> http://matplotlib.sourceforge.net/search.html?q=codex+baseline

Thanks, that will be very useful.  Actually the page I was looking for was:

http://matplotlib.sourceforge.net/examples/pylab_examples/demo_text_rotation_mode.html?highlight=codex%20baseline

So far,
Friedrich

Eric,

thanks for follow-up! No, I didn't miss it.  Saw it Sunday night my time and spent time with my son getting him ready for start of university istead.

Re the setup of Python ... I will do that this week. 

Here's the code extract:

#!/usr/bin/env python
# −*− coding: UTF−8 −*−

import matplotlib
matplotlib.use('Agg') # before importing pyplot per docs
import matplotlib.pyplot as plt
import numpy as np

[snipped the import of Python and Django libraries]

###########################################################
def CreateAllWebSite():
    # whichever of these is called first is done fine.  Second flawed.  calling one at at time, each works ok
    CreateMemberStatCategoryFigure()
    CreateMemberStatFigure()
###############################################
def CreateMemberStatFigure():
    """
    This function creates the member statistics bar chart that is published multiple locations.
    """
    membertypes=Membertype.objects.filter(active=True).order_by('emailorder').exclude(typecode='EXEC')
    mtypes=[]
    barchartcolor=[]
    mtypecodes=[]
    for i in membertypes:
        mtypecodes.append(i.typecode)
        mtypes.append(i.description)
        barchartcolor.append(i.barchartcolor)
    memberstatsm=MemberStatistics.monthly_objects.all()
    memberstatsp=MemberStatistics.monthlypivotbytype_objects.all()
    for i in memberstatsm:
        monthend=i['Date']

##### this section in development. a more generic way; but can't get the calls to p=plt to work. 
##### proably needs version 1.0 of Matplotlib or something.  As it works, optimise and fix later.
#    print mtypes,mtypecodes,barchartcolor    
#    width=24
#    month=[]
#    for j in memberstatsp:
#        month.append(None2Zero(j['month']))
#    N=len(month)
#    base=np.zeros(N,int)    # start off the base with zeros
#    for i in membertypes:
#        array1=[]
#        for j in memberstatsp:
#            array1.append(None2Zero(j[i.typecode]))
#        array1=np.array(array1)
#        p=plt.bar(month,array1,width,bottom=base,linewidth=0)
#        base=base+array1
#    plt.yticks(np.arange(0,200,20))
#    plt.title(ORG_NAME+' Membership by Month')
#    plt.ylabel('Count')
#    plt.show()
#    return
    
#####        
        

    corp1=[]
    corp2=[]
    corp3=[]
    ordm=[]
    ordr=[]
    ordh=[]
    month=[]
    for i in memberstatsp:
        month.append(None2Zero(i['month']))
        corp1.append(None2Zero(i['CORP1']))
        corp2.append(None2Zero(i['CORP2']))
        corp3.append(None2Zero(i['CORP3']))
        ordm.append(None2Zero(i['ORM']))
        ordr.append(None2Zero(i['ORMR']))
        ordh.append(None2Zero(i['HON']))
    N=len(month)
    corp1=np.array(corp1)
    corp2=np.array(corp2)
    corp3=np.array(corp3)
    ordm=np.array(ordm)
    ordr=np.array(ordr)
    ordh=np.array(ordh)

    base=np.zeros(N,int)
    base1=base+ordh
    base2=base1+corp3
    base3=base2+corp2
    base4=base3+corp1
    base5=base4+ordm
    
    width = 24        # the width of the bars: can also be len(x) sequence
    
    p1 = plt.bar(month, ordh, width, color=barchartcolor[0],linewidth=0)
    p2 = plt.bar(month, corp3, width, color=barchartcolor[1],bottom=base1,linewidth=0)
    p3 = plt.bar(month, corp2, width, color=barchartcolor[2],bottom=base2,linewidth=0)
    p4 = plt.bar(month, corp1, width, color=barchartcolor[3],bottom=base3,linewidth=0)
    p5 = plt.bar(month, ordm, width, color=barchartcolor[4],bottom=base4,linewidth=0)
    p6 = plt.bar(month, ordr, width, color=barchartcolor[5],bottom=base5,linewidth=0)

    plt.yticks(np.arange(0,200,20))
    plt.title(ORG_NAME+' Membership by Month')
    plt.ylabel('Count')
    plt.legend( (p1[0], p2[0], p3[0], p4[0], p5[0], p6[0]),mtypes,loc='best') 
    plt.show()
    
    matplotlib.rcParams['figure.figsize']=(3,2)
    
    fn=TEMP_DIR+SOC_MEMBER_STAT_BARCHART_FIGURE
    print "    CreateMemberStatFigure(): Saving ",fn,"..."
    plt.savefig(fn)
    
    return

###############################################
def CreateMemberStatCategoryFigure():
    """
    This function creates the member statistics category bar chart that is published multiple locations.
    """
    dt=strftime("%d %B %Y", gmtime())
    membercategories=Membercategory.objects.all()
    mcats=[]
    cnt=[]
    for i in membercategories:
        mcats.append(i.category)
        cnt.append(Member.Active_objects.filter(membercategory__category=i.category).count())    
    xlocations = np.array(range(len(mcats)))+0.5
    
    print mcats
    print cnt
    p1 = plt.bar(xlocations,cnt)
    plt.figtext(0.15,0.85,dt)
    plt.xticks(xlocations+0.4, mcats,rotation=17)
    plt.title(ORG_NAME+' Membership by Member Category')
    plt.ylabel('Count')
    plt.show()
    
    matplotlib.rcParams['figure.figsize']=(3,2)
    
    fn=TEMP_DIR+SOC_MEMBER_STAT_BARCHART_CAT_FIGURE
    print "    CreateMemberStatCategoryFigure(): Saving ",fn,"..."
    plt.savefig(fn)
    print "    CreateMemberStatCategoryFigure(): ",SOC_MEMBER_STAT_BARCHART_CAT_FIGURE," created ..."
    
    return

################################################



On 9 Aug 2010, at 07:41, Eric Firing wrote:

> Did you miss my reply to the list?
> 
> Eric
> 
> -------- Original Message --------
> Subject: Re: [Matplotlib-users] Unable to make sequential calls to functions with create matplotlib graphs ... second gets jumbled
> Date: Sun, 08 Aug 2010 08:10:37 -1000
> From: Eric Firing <ef...@ha...>
> Organization: University of Hawaii
> To: mat...@li...
> 
> On 08/08/2010 12:53 AM, Rob Schneider wrote:
>> Running 0.98.5.3
>> (would upgrade to 1.0 but having trouble installing it on Mac OS X as it doesn't appear to like the version of Python Apple provides).
>> 
> 
> Standard procedure on the Mac is to install the python from python.org.
> That way you don't risk fouling up the python that OS X is depending on,
> and at the same time your own code can use the version of python that
> most extension packages are built for.
> 
> 
>> I have a module where there are two functions which call matplotlib to each create a different graph which both saved to an output PNG file.  The import statements:
>> 
>> import matplotlib
>> matplotlib.use('Agg')
>> import matplotlib.pyplot as plt
>> import numpy as np
>> 
>> are at the top of the module file.  Neither function imports anything specifically, relying on the above imports.
>> 
>> I call each function sequentially.  The result is that the first graph is created correctly. The second graph is flawed, and in fact shows the legend from the first function.  When I reverse the function calls, which ever graphing function is called first is ok, but second is flawed. If I call each function on it's own, within the same Python session, the graphs (both) get produced flawlessly.
>> 
>> both charts are bar charts; but one does a stacked bar and the other is a simple bar chart.
>> 
>> Why would I not be able to call the two functions sequentially?  Why does the first call interfere with the second call?
>> 
> 
> I don't think anyone is going to be able to answer this question until
> you provide a minimal stand-alone code example.  Also, the process of
> stripping your code down to the shortest example that displays the
> undesired behavior may show you where the problem lies.
> 
> Eric
> 

Hello,
I recently upgraded to Enthought python 2.6 an so I have now matplotlib 
0.99.3
My scripts now no longer seem to find the pyGTK since when I run them I 
got this message:

Traceback (most recent call last):
   File "./sed_plotb_v2.py", line 14, in <module>
     from pylab import *
   File 
"/Library/Frameworks/Python.framework/Versions/6.2/lib/python2.6/site-packages/pylab.py", 
line 1, in <module>
     from matplotlib.pylab import *
   File 
"/Library/Frameworks/Python.framework/Versions/6.2/lib/python2.6/site-packages/matplotlib/pylab.py", 
line 247, in <module>
     from matplotlib.pyplot import *
   File 
"/Library/Frameworks/Python.framework/Versions/6.2/lib/python2.6/site-packages/matplotlib/pyplot.py", 
line 78, in <module>
     new_figure_manager, draw_if_interactive, show = pylab_setup()
   File 
"/Library/Frameworks/Python.framework/Versions/6.2/lib/python2.6/site-packages/matplotlib/backends/__init__.py", 
line 25, in pylab_setup
     globals(),locals(),[backend_name])
   File 
"/Library/Frameworks/Python.framework/Versions/6.2/lib/python2.6/site-packages/matplotlib/backends/backend_gtkagg.py", 
line 10, in <module>
     from matplotlib.backends.backend_gtk import gtk, FigureManagerGTK, 
FigureCanvasGTK,\
   File 
"/Library/Frameworks/Python.framework/Versions/6.2/lib/python2.6/site-packages/matplotlib/backends/backend_gtk.py", 
line 11, in <module>
     raise ImportError("Gtk* backend requires pygtk to be installed.")
ImportError: Gtk* backend requires pygtk to be installed.


is any way to help matplotlib to point the where pyGTK is or go around 
the problem with another backend?

Thanks,
Nino

-- 

Antonino Cucchiara, PhD
Postdoctoral Scholar
Lawrence Berkeley National Lab
UC Berkeley
website: www.astro.psu.edu/~cucchiara/

Hello,
I am running a MacOS Snow leopard and I recently istalled the python 2.6 
via Enthought.
I wanted to upgrade the matplotlib 1.0 but when I run the .mpkg the 
installation don't start because
it seems not recognizing the python 2.6 is installed (I do have now 2.6.6).

Any suggestions?

Thanks,
Nino

-- 

Antonino Cucchiara, PhD
Postdoctoral Scholar
Lawrence Berkeley National Lab
UC Berkeley
website: www.astro.psu.edu/~cucchiara/

On Sun, Aug 8, 2010 at 8:03 PM, Friedrich Romstedt
<fri...@gm...> wrote:

> Yes, I agree fully, but have been busy the past few days :-)
>
> Friedrich
>
> P.S.: You can also try
> http://matplotlib.sourceforge.net/examples/pylab_examples/alignment_test.html
> , I believe there was some other manual around also covering the
> baseline option, but I cannot find it now.
>

The "codex" option on the search page helps with this; you can easily
search for "codex baseline"::

  * the default  search page: http://matplotlib.sourceforge.net/search.html

  * what is codex?:
http://matplotlib.sourceforge.net/faq/howto_faq.html#search-examples

  * an example using "codex baseline":
http://matplotlib.sourceforge.net/search.html?q=codex+baseline

Basically, you go to the default mpl homepage or any sub-page, and
from there, click the "Search" link, and then enter "codex YOURPHRASE"
where codex stands for "code example" and will return all code
examples containing your phrase.

JDH

Hi,

On my Macbook Pro I use Python 2.6 as provided by the Enthought Python  
Distribution. I ran into some problem with Axes3D so I decided to  
upgrade to the latest version from source. While trying that I got a  
similiar error message as discussed in

http://www.mail-archive.com/mat...@li.../msg12938.html

i.e. error: png.h: No such file or directory

In short matplotlib could not find libpng (or png.h from that). I had  
libpng installed via fink in the usual place /sw/include etc. This  
seems to be a quite natural choice, so I wondered why matplotlib does  
not find that out of the box. Following the tips in the above post I  
looked at basedir in setupext.py For some reason the basedirs for  
darwin where commented out. Uncommenting the basedirs did the trick  
for me [1] and matplotlib happily installed with the libpng from fink.

Hope this will help other mac users. Thanks for the great work on  
matplotlib!

Cheers,

Markus



[1] Here the relevant lines for setupext.py

     'darwin' : ['/sw/lib/freetype2', '/sw/lib/freetype219', '/usr/ 
local',
                 '/usr', '/sw'],
     # it appears builds with darwin are broken because of all the
     # different flags the deps can be compile with, so I am pushing
     # people to :
     #   make -f make.osx fetch deps mpl_build mpl_install

     #'_darwin' : [],

On Sun, Aug 8, 2010 at 10:38 PM, Simon Friedberger
<sim...@a-...> wrote:
> a) I tried using the twinx() command like in that example but the axes
> seem to be scaled differently i.e. the ticks on the right don't have the
> same spacing as on the left.
>

twinx() creates an axes whose y-axis is independent from the original axes.

> b) Is it necessary to do that with twinx() or is there a way to set the
> labels for both sides independently?
>

Within a same axes, you cannot change the ticklabels of one side
without affecting the other side. This is also true between two axes
with shared axis.

One option would be to sync the limits of two axis manually (or
automatically using event). Another option you may consider is to use
the axes_grid1 toolkit (assuming that you're using mpl v1.0).



    from mpl_toolkits.axes_grid1.parasite_axes import SubplotHost

    fig = plt.figure()
    ax = SubplotHost(fig, 111)
    fig.add_subplot(ax)

    ax2 = ax.twin() # "twin", not "twinx"
    ax2.axis["top"].toggle(ticklabels=False) # make ticklabels at top invisible

    ax2.set_yticks([0.,0.5, 1.])
    ax2.set_yticklabels(["A", "B", "C"])

-JJ

It and scipy work very well. I have 64-bit Win 7, HP Pavilon desktop. A 
re-examination shows the install files are all exe files, except I think 
for Python. I'm using 2.5 and need to stay there along with numpy 1.2.0. 
When you say uninstall, you must mean delete the file under C:\Python. 
Well, it's really not me that needs the help, it's my partner. He went 
off to 1.4, and it has caused trouble. Let's not go there. :-)

On 8/8/2010 4:33 PM, Stephen George wrote:
>    On 9/08/2010 9:19 AM, Wayne Watson wrote:
>    
>> See Subject. I use matplotlib, scipy, numpy and possibly one other
>> module. If I go to the control panel, I only see numpy listed. Why? I
>> use a search and find only numpy and Python itself. How can matplotlib
>> and scipy be uninstalled?
>>
>>      
> In my Win7 (64bit) I have 32 bit python 2.6 installed.
>
> matplotlib and numpy show under the add remove program functionality
>
> "Python 2.6 matplotlib-0.99.3"
> "Python 2.6 numpy-1.4.1"
>
> Along with heaps of other libraries I have installed
>
> Are you sure matplotlib is installed AND working?
>
> To remove I can only suggest you install again and then remove
>
> Steve
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by
>
> Make an app they can't live without
> Enter the BlackBerry Developer Challenge
> http://p.sf.net/sfu/RIM-dev2dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>    

-- 
            Wayne Watson (Watson Adventures, Prop., Nevada City, CA)

              (121.015 Deg. W, 39.262 Deg. N) GMT-8 hr std. time)
               Obz Site:  39° 15' 7" N, 121° 2' 32" W, 2700 feet

             "An experiment is a question which science poses to
              Nature, and a measurement is the recording of
              Nature’s answer." -- Max Planck


                     Web Page:<www.speckledwithstars.net/>

2010/8/8 Simon Friedberger <sim...@a-...>:
> I have found a solution. I'm not sure if it's good or intended but the
> following works:
>
> for label in xax.get_ticklabels():
>        label.set_rotation(45)
>        label.set_horizontalalignment('left')

This is fully intended.  Maybe you can compare with the version
attached, I believe my version should have some minor improvement (the
labels are anchored all with the baseline's leftmost point at the
center of the grid cell).  I have not tried your version, though.

You got rid of the +0.4 with your solution I believe?

Also, for the previous pocoo version still online there, I had a
problem with the "superduperverylonglabel".  It was shiftet to the
left.  What is understandable from the point of view that matplotlib
anchored the labels' center point at x + 0.4.

> Please comment. Apart from that it's here for people who look for a
> solution to the same question.

Yes, I agree fully, but have been busy the past few days :-)

Friedrich

P.S.: You can also try
http://matplotlib.sourceforge.net/examples/pylab_examples/alignment_test.html
, I believe there was some other manual around also covering the
baseline option, but I cannot find it now.

  On 9/08/2010 9:19 AM, Wayne Watson wrote:
> See Subject. I use matplotlib, scipy, numpy and possibly one other
> module. If I go to the control panel, I only see numpy listed. Why? I
> use a search and find only numpy and Python itself. How can matplotlib
> and scipy be uninstalled?
>
In my Win7 (64bit) I have 32 bit python 2.6 installed.

matplotlib and numpy show under the add remove program functionality

"Python 2.6 matplotlib-0.99.3"
"Python 2.6 numpy-1.4.1"

Along with heaps of other libraries I have installed

Are you sure matplotlib is installed AND working?

To remove I can only suggest you install again and then remove

Steve

See Subject. I use matplotlib, scipy, numpy and possibly one other 
module. If I go to the control panel, I only see numpy listed. Why? I 
use a search and find only numpy and Python itself. How can matplotlib 
and scipy be uninstalled?

-- 
            Wayne Watson (Watson Adventures, Prop., Nevada City, CA)

              (121.015 Deg. W, 39.262 Deg. N) GMT-8 hr std. time)
               Obz Site:  39° 15' 7" N, 121° 2' 32" W, 2700 feet

             "An experiment is a question which science poses to
              Nature, and a measurement is the recording of
              Nature’s answer." -- Max Planck


                     Web Page:<www.speckledwithstars.net/>

Eric Firing <efiring@...> writes:

> Try "ipython -pylab".  We don't want the threaded version.
> 
> Eric

Well, it seems matplotlib svn is required. I got it working after I upgraded
from 1.0 to svn (revision 8624 to be precise) and with Ipython 0.11.alpha1.
Thanks for the tip again, I am a happier man now ;)

Jorge

On 08/08/2010 12:53 AM, Rob Schneider wrote:
> Running 0.98.5.3
> (would upgrade to 1.0 but having trouble installing it on Mac OS X as it doesn't appear to like the version of Python Apple provides).
>

Standard procedure on the Mac is to install the python from python.org. 
That way you don't risk fouling up the python that OS X is depending on, 
and at the same time your own code can use the version of python that 
most extension packages are built for.


> I have a module where there are two functions which call matplotlib to each create a different graph which both saved to an output PNG file.  The import statements:
>
> import matplotlib
> matplotlib.use('Agg')
> import matplotlib.pyplot as plt
> import numpy as np
>
> are at the top of the module file.  Neither function imports anything specifically, relying on the above imports.
>
> I call each function sequentially.  The result is that the first graph is created correctly. The second graph is flawed, and in fact shows the legend from the first function.  When I reverse the function calls, which ever graphing function is called first is ok, but second is flawed. If I call each function on it's own, within the same Python session, the graphs (both) get produced flawlessly.
>
> both charts are bar charts; but one does a stacked bar and the other is a simple bar chart.
>
> Why would I not be able to call the two functions sequentially?  Why does the first call interfere with the second call?
>

I don't think anyone is going to be able to answer this question until 
you provide a minimal stand-alone code example.  Also, the process of 
stripping your code down to the shortest example that displays the 
undesired behavior may show you where the problem lies.

Eric

Hi list members!

I'm not sure if this is bad style or anything but I'm trying to do
something like in this example:
http://matplotlib.sourceforge.net/examples/pylab_examples/barchart_demo2.html

I would like to haver different labels on both sides.
a) I tried using the twinx() command like in that example but the axes
seem to be scaled differently i.e. the ticks on the right don't have the
same spacing as on the left.

Sample code is here:
http://paste.pocoo.org/show/247636/

b) Is it necessary to do that with twinx() or is there a way to set the
labels for both sides independently?

Best
Simon

Running 0.98.5.3
(would upgrade to 1.0 but having trouble installing it on Mac OS X as it doesn't appear to like the version of Python Apple provides).

I have a module where there are two functions which call matplotlib to each create a different graph which both saved to an output PNG file.  The import statements:

import matplotlib
matplotlib.use('Agg') 
import matplotlib.pyplot as plt
import numpy as np

are at the top of the module file.  Neither function imports anything specifically, relying on the above imports.

I call each function sequentially.  The result is that the first graph is created correctly. The second graph is flawed, and in fact shows the legend from the first function.  When I reverse the function calls, which ever graphing function is called first is ok, but second is flawed. If I call each function on it's own, within the same Python session, the graphs (both) get produced flawlessly.

both charts are bar charts; but one does a stacked bar and the other is a simple bar chart.

Why would I not be able to call the two functions sequentially?  Why does the first call interfere with the second call?

--rms

Running 0.98.5.3
(would upgrade to 1.0 but having trouble installing it on Mac OS X as it doesn't appear to like the version of Python Apple provides).

I have a module where there are two functions which call matplotlib to each create a different graph which both saved to an output PNG file.  The import statements:

import matplotlib
matplotlib.use('Agg') 
import matplotlib.pyplot as plt
import numpy as np

are at the top of the module file.  Neither function imports anything specifically, relying on the above imports.

I call each function sequentially.  The result is that the first graph is created correctly. The second graph is flawed, and in fact shows the legend from the first function.  When I reverse the function calls, which ever graphing function is called first is ok, but second is flawed. If I call each function on it's own, within the same Python session, the graphs (both) get produced flawlessly.

both charts are bar charts; but one does a stacked bar and the other is a simple bar chart.

Why would I not be able to call the two functions sequentially?  Why does the first call interfere with the second call?

--rms

On 16:00 Fri 06.08.10, Simon Friedberger wrote:
> It is about the positioning of the rotated labels.
> The code is here:
> http://paste.pocoo.org/show/246870/
> 
> Note that in line 36 I had already remarked about the hack I used. Now I
> noticed that if the labels have different lengths rotating them gives
> different positions. What I really want to do is rotate about the bottom
> of the labels. Can that be done?

I have found a solution. I'm not sure if it's good or intended but the
following works:

for label in xax.get_ticklabels():
	label.set_rotation(45)
	label.set_horizontalalignment('left')

Please comment. Apart from that it's here for people who look for a
solution to the same question.

Best
Simon

On 08/07/2010 11:44 AM, Jorge Scandaliaris wrote:
> Eric Firing<efiring@...>  writes:
>> Running mpl from svn, with gtkagg backend, on ubuntu, but with ipython
>> 0.11alpha1, I can't reproduce the problem; your example works for me.  I
>> suspect the difference is that ipython 0.11 uses a non-threaded
>> interactive mode, while 0.10 uses the threaded mode.  Is updating your
>> ipython an option for you?
>>
>> Eric
>>
>
> Well, I just installed ipython 0.11.alpha1.git and my example doesn't work at
> all (first plot is displayed, but neither mouse nor key presses have any
> effect). Could it be that I use mpl.1.0 instead of svn? How do you start
> ipython? I usually do "ipython -gthread -pylab", but maybe that is also
> contributing to the problem.

Try "ipython -pylab".  We don't want the threaded version.

Eric


>
> Regards,
>
> Jorge
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by
>
> Make an app they can't live without
> Enter the BlackBerry Developer Challenge
> http://p.sf.net/sfu/RIM-dev2dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Eric Firing <efiring@...> writes:
> Running mpl from svn, with gtkagg backend, on ubuntu, but with ipython 
> 0.11alpha1, I can't reproduce the problem; your example works for me.  I 
> suspect the difference is that ipython 0.11 uses a non-threaded 
> interactive mode, while 0.10 uses the threaded mode.  Is updating your 
> ipython an option for you?
> 
> Eric
> 

Well, I just installed ipython 0.11.alpha1.git and my example doesn't work at
all (first plot is displayed, but neither mouse nor key presses have any
effect). Could it be that I use mpl.1.0 instead of svn? How do you start
ipython? I usually do "ipython -gthread -pylab", but maybe that is also
contributing to the problem.

Regards,

Jorge

Eric Firing <efiring@...> writes:


> Running mpl from svn, with gtkagg backend, on ubuntu, but with ipython 
> 0.11alpha1, I can't reproduce the problem; your example works for me.  I 
> suspect the difference is that ipython 0.11 uses a non-threaded 
> interactive mode, while 0.10 uses the threaded mode.  Is updating your 
> ipython an option for you?
> 
> Eric
> 

Yes, it is. I'll give it a try. Thanks for the info.

Jorge

On 08/05/2010 10:54 PM, Jorge Scandaliaris wrote:
> Hi,
> Looking for a way to analyze a set of images I found waitforbuttonpress, which
> basically does what I need: allows me to display a few plots/scatter/etc for as
> long as I need; then when pressing a key move to the next image and redo all
> plots/scatter/etc.
> Now my problem is that waitforbuttonpress (or the (mis)use I gave it) forces me
> to close the ipython session frequently and start over again. This is at least
> inconvenient, and a few times forced me to recalculate things that took some
> time (and that I had forgot to save).
> I know this might be backend-related. I use GTKAgg, on ubuntu 10.04, mpl-1.0.0
> (locally built). Below you'll find a script I created where I see this behavior,
> the ipython console's output, and the steps I followed.
>
> Steps followed and description:
> 1- Run script, click mouse a few times (plot is closed and redrawn), finally
> press a key. The plot is closed, and the console returns to be active. This
> corresponds to "In [1]" in the ipython console's output
> 2- Trying to run anything at the ipython console results in a looong delay and a
> message like in "In [2]" at the console. The console works fine after this,
> though, at least with non matplotlib-related stuff
> 3- Running the script again results in the output shown in "In [3]". The script
> seems to runs fine, i.e. clicks cause the redrawing of the plot, though pressing
> any key (which should break the loop and end the script) will end the script but
> won't close the plot windows. The console works mostly ok, the only way to close
> the plot window and ipython is by killing the ipython process.
> 4- In some other cases, I ended up with an unresponsive console that required
> killing the ipython process as well.
> This test case is not comprehensive of the things I've experienced, but it might
> show a possible cause of this behavior.

Jorge,

Running mpl from svn, with gtkagg backend, on ubuntu, but with ipython 
0.11alpha1, I can't reproduce the problem; your example works for me.  I 
suspect the difference is that ipython 0.11 uses a non-threaded 
interactive mode, while 0.10 uses the threaded mode.  Is updating your 
ipython an option for you?

Eric

>
> Regards,
>
> Jorge
>
> ------------------------
> Ipython console's output:
>
> In [1]: run doct/intrinsic-images/test_waitforbuttonpress.py
> /home/jscandal/lib/python2.6/site-packages/matplotlib/backend_bases.py:2051:
> DeprecationWarning: Using default event loop until function specific to this GUI
> is implemented
>    warnings.warn(str,DeprecationWarning)
>
> In [2]: run doct/intrinsic-images/test_waitforbuttonpress.py
> Warning: Timeout for mainloop thread exceeded
> switching to nonthreaded mode (until mainloop wakes up again)
>
> In [3]: run doct/intrinsic-images/test_waitforbuttonpress.py
> ---------------------------------------------------------------------------
> ValueError                                Traceback (most recent call last)
>
> /usr/lib/pymodules/python2.6/IPython/Shell.pyc in on_timer(self)
>      847
>      848         update_tk(self.tk)
> -->  849         self.IP.runcode()
>      850         time.sleep(0.01)
>      851         return True
>
> /usr/lib/pymodules/python2.6/IPython/Shell.pyc in runcode(self)
>      455         # code modifies it, we restore our own handling.
>
>      456         try:
> -->  457             signal(SIGINT,sigint_handler)
>      458         except SystemError:
>      459             # This happens under Windows, which seems to have all sorts
>
>
> ValueError: signal only works in main thread
> -------------------------
>
> script used in the test:
> ------------------------
> import numpy as np
> import matplotlib as mpl
> import matplotlib.pyplot as plt
>
> data = np.random.rand(10,2)
>
> while True:
>      fig, ax = plt.subplots(1,1)
>      ax.scatter(data.T[0], data.T[1])
>      plt.show()
>      if fig.waitforbuttonpress(timeout=-1):
> 	plt.close('all')
> 	break
>      plt.close('all')
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by
>
> Make an app they can't live without
> Enter the BlackBerry Developer Challenge
> http://p.sf.net/sfu/RIM-dev2dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users