matplotlib-users Mailing List for matplotlib

Hi Goyo,
>> That is correct. If you have a call to show(), then the script should not
>> finish on their own until the windows are closed -- regardless of whether or
>> not "interactive" is True or False. The "interactive" setting should only
>> dictate whether or not the script execution pauses or not at the call to
>> show().
> 
> Then the script is supposed to keep itself alive, after executing the
> last statment, until the plot windows are closed? Does not work that
> way for me (tkagg, qt4agg and gtk*) and I wouldn't expect that.
Yes, this is how things used to work on my mac (1 week ago). After upgrade, no longer works this way. Reverting to old compilation - still no longer works.
> BTW this may be better than using raw_input:
> 
> import matplotlib.pyplot as plt
> plt.ion()
> plt.plot([1.6, 2.7]) # The plot windows shows up.
> # Do stuff, even user interaction, more plots, etc.
> # ...
> # Wait until all plot windows are closed.
> plt.ioff()
> plt.show()
This reproduces (as far as I can tell) the exact behavior I used to have.
Interesting to note. If I have "interactive : True" in my matplotlibrc file, and then 
import matplotlib.pyplot as plt
# do everything I used to do
# including call functions that do other plots etc.
plt.ioff()
plt.show()
this also produces the above behavior (what used to work).
So, if I am in interactive mode, then matplotlibrc opens and closes the figures for me. However, if after making all my plots, just before calling plt.show(), I turn off interactive mode, then the plots stay open.
This does not follow the logic I would expect - ie interactive means I would have to close the figures interactively. But it gives the behavior I have come to expect.
Now back to my other problems (I'll start a new thread).
Cheers,
Andre

2012年6月19日 Benjamin Root <ben...@ou...>:
>
>
> On Tue, Jun 19, 2012 at 2:40 PM, Andre' Walker-Loud <wal...@gm...>
> wrote:
>>
>> Hi Goyo,
>>
>> > 2012年6月19日 Andre' Walker-Loud <wal...@gm...>:
>> >> But my plots still vanish as soon as the script is done :
>> >
>> > That's to be expected. You can make the script not to end until the
>> > user ask for it explicitly:
>> >
>> > raw_input('Press <Enter> when you are done')
>>
>> If this is expected - it is a new feature.
>>
>> My understanding was that changing
>>
>> interactive : True
>>
>> in the matplotlibrc file, then the plots would not vanish until explicitly
>> closed by the user.
>>
>> Is my understanding incorrect?
>>
>>
>> Thanks,
>>
>> Andre
>
>
> That is correct. If you have a call to show(), then the script should not
> finish on their own until the windows are closed -- regardless of whether or
> not "interactive" is True or False. The "interactive" setting should only
> dictate whether or not the script execution pauses or not at the call to
> show().
Then the script is supposed to keep itself alive, after executing the
last statment, until the plot windows are closed? Does not work that
way for me (tkagg, qt4agg and gtk*) and I wouldn't expect that.
BTW this may be better than using raw_input:
import matplotlib.pyplot as plt
plt.ion()
plt.plot([1.6, 2.7]) # The plot windows shows up.
# Do stuff, even user interaction, more plots, etc.
# ...
# Wait until all plot windows are closed.
plt.ioff()
plt.show()
Goyo

Hi Goyo,
I see! I am using the macosx backend. That was the hint I needed. Thanks a lot!
Maximilian
On Tue, Jun 19, 2012 at 9:27 PM, Goyo <goy...@gm...> wrote:
> 2012年6月8日 Maximilian Fabricius <mfa...@gm...>:
>> Hi all,
>>
>> it seems that whenever I plot something, a window opens.
>>
>> from matplotlib import pylab
>> import numpy
>> pylab.plot(numpy.random.normal(size=100))
>>
>> Now, I have code that is supposed to produce diagnostic plots as PDFs. Only
>> when I pass a command line option the script should actually open a
>> window and also display
>> the results on the screen.
>>
>> I am pretty sure that in earlier times the window would only open when
>> I call pylab.show().
>>
>> It may well be that I messed too much with my matplotlibrc, if you
>> could just point me in the right direction ...
>>
>> I use matplotlib 1.0.1 on a Mac with the MacOSX backend.
>
> From http://matplotlib.sourceforge.net/faq/usage_faq.html#non-interactive-example:
>
> ----------------------------------------------------------
> import matplotlib.pyplot as plt
> plt.ioff()
> plt.plot([1.6, 2.7])
>
> Nothing happened–or at least nothing has shown up on the screen
> (unless you are using macosx backend, which is anomalous).
> ----------------------------------------------------------
>
>
> Goyo
-- 
--------------------------------------------------------------------
 Maximilian Fabricius
 * Max Planck Institute for Extraterrestrial Physics (MPE)
   Giessenbachstrasse, D-85748 Garching, Germany
 * University Observatory Munich /  Wendelstein Observatory
   Scheinerstr. 1, D-81679 München, Germany
 eMail: mx...@mp...
 Phone: +49 89 30000 3779
 --------------------------------------------------------------------

>> If this is expected - it is a new feature.
>> 
>> My understanding was that changing
>> 
>> interactive : True
>> 
>> in the matplotlibrc file, then the plots would not vanish until explicitly closed by the user.
>> 
>> Is my understanding incorrect?
>> 
>> 
>> Thanks,
>> 
>> Andre
>> 
>> That is correct. If you have a call to show(), then the script should not finish on their own until the windows are closed -- regardless of whether or not "interactive" is True or False. The "interactive" setting should only dictate whether or not the script execution pauses or not at the call to show(). Have you tried a different backend as a temporary work-around such as QTAgg, QT4Agg, GTKAgg, TkAgg?
> 
> This behavior is observed with both TkAgg and MacOSX.
> I don't have pygtk or pyqt installed, so couldn't test the others.
In case it matters, I am using the i386 installation and not the x86_64.
I have more info - which may be helpful.
Inspired by a related thread - which Goyo just answered, I tried the following
>>> import matplotlib.pyplot as plt
>>> plt.plot([1.6, 2.7])
In an interactive python session (also in ipython), this produced the plot, and it stayed up until I closed it.
I converted it to a little script
===
#!/Library/Frameworks/Python.framework/Versions/Current/bin/python 
import matplotlib.pyplot as plt
plt.plot([1.6, 2.7])
print plt.get_backend()
plt.show()
===
this prints to screen "TkAgg", but the plot disappears as soon as the script finishes.
Thanks,
Andre

2012年6月8日 Maximilian Fabricius <mfa...@gm...>:
> Hi all,
>
> it seems that whenever I plot something, a window opens.
>
> from matplotlib import pylab
> import numpy
> pylab.plot(numpy.random.normal(size=100))
>
> Now, I have code that is supposed to produce diagnostic plots as PDFs. Only
> when I pass a command line option the script should actually open a
> window and also display
> the results on the screen.
>
> I am pretty sure that in earlier times the window would only open when
> I call pylab.show().
>
> It may well be that I messed too much with my matplotlibrc, if you
> could just point me in the right direction ...
>
> I use matplotlib 1.0.1 on a Mac with the MacOSX backend.
>From http://matplotlib.sourceforge.net/faq/usage_faq.html#non-interactive-example:
----------------------------------------------------------
import matplotlib.pyplot as plt
plt.ioff()
plt.plot([1.6, 2.7])
Nothing happened–or at least nothing has shown up on the screen
(unless you are using macosx backend, which is anomalous).
----------------------------------------------------------
Goyo

Hi Ben,
> On Jun 19, 2012, at 12:15 PM, Benjamin Root wrote:
> On Tue, Jun 19, 2012 at 2:40 PM, Andre' Walker-Loud <wal...@gm...> wrote:
> Hi Goyo,
> 
> > 2012年6月19日 Andre' Walker-Loud <wal...@gm...>:
> >> But my plots still vanish as soon as the script is done :
> >
> > That's to be expected. You can make the script not to end until the
> > user ask for it explicitly:
> >
> > raw_input('Press <Enter> when you are done')
> 
> If this is expected - it is a new feature.
> 
> My understanding was that changing
> 
> interactive : True
> 
> in the matplotlibrc file, then the plots would not vanish until explicitly closed by the user.
> 
> Is my understanding incorrect?
> 
> 
> Thanks,
> 
> Andre
> 
> That is correct. If you have a call to show(), then the script should not finish on their own until the windows are closed -- regardless of whether or not "interactive" is True or False. The "interactive" setting should only dictate whether or not the script execution pauses or not at the call to show(). Have you tried a different backend as a temporary work-around such as QTAgg, QT4Agg, GTKAgg, TkAgg?
This behavior is observed with both TkAgg and MacOSX.
I don't have pygtk or pyqt installed, so couldn't test the others.
Andre

On Tue, Jun 19, 2012 at 2:40 PM, Andre' Walker-Loud <wal...@gm...>wrote:
> Hi Goyo,
>
> > 2012年6月19日 Andre' Walker-Loud <wal...@gm...>:
> >> But my plots still vanish as soon as the script is done :
> >
> > That's to be expected. You can make the script not to end until the
> > user ask for it explicitly:
> >
> > raw_input('Press <Enter> when you are done')
>
> If this is expected - it is a new feature.
>
> My understanding was that changing
>
> interactive : True
>
> in the matplotlibrc file, then the plots would not vanish until explicitly
> closed by the user.
>
> Is my understanding incorrect?
>
>
> Thanks,
>
> Andre
That is correct. If you have a call to show(), then the script should not
finish on their own until the windows are closed -- regardless of whether
or not "interactive" is True or False. The "interactive" setting should
only dictate whether or not the script execution pauses or not at the call
to show(). Have you tried a different backend as a temporary work-around
such as QTAgg, QT4Agg, GTKAgg, TkAgg?
Ben Root

Hi Goyo,
> 2012年6月19日 Andre' Walker-Loud <wal...@gm...>:
>> But my plots still vanish as soon as the script is done :
> 
> That's to be expected. You can make the script not to end until the
> user ask for it explicitly:
> 
> raw_input('Press <Enter> when you are done')
If this is expected - it is a new feature.
My understanding was that changing 
interactive : True
in the matplotlibrc file, then the plots would not vanish until explicitly closed by the user.
Is my understanding incorrect?
Thanks,
Andre

2012年6月19日 Andre' Walker-Loud <wal...@gm...>:
> But my plots still vanish as soon as the script is done :
That's to be expected. You can make the script not to end until the
user ask for it explicitly:
raw_input('Press <Enter> when you are done')
Goyo

2012年6月19日 Goyo <goy...@gm...>:
>> I think it normally shouldn't give the object ID.
>
> Yes, it should.
>
Sorry, not an "object ID" but a string representation of the returned object.

2012年6月19日 Mogliii <mo...@gm...>:
> Hi,
>
> Kubuntu 12.04 with python 2.7 and matplotlib 1.1.1
>
> Suddenly the matplotlib window does not open anymore. I tried from
> konsole with pyton, ipython and also from eclipse with pydev. Last time
> I know it worked was maybe 5 days ago, haven't tried since then.
>
> In ipython --pylab it looks like this:
>
> In [1]: plt.plot(np.arange(5))
> Out[1]: [<matplotlib.lines.Line2D at 0x2804fd0>]
>
> In [2]: plt.show()
>
> In [3]:
>
> but no widow opens.
Make sure an interactive backend is in use:
plt.get_backend()
>
> I think it normally shouldn't give the object ID.
Yes, it should.
Goyo

Hi All,
Sort of hijacking my own thread here. I posted another issue yesterday or the day before, and this has caused me to find some bigger problems...
> Most likely, there was a change to your matplotlibrc file. There is a setting in there for "interactive" and by default, it is set to False. If you uncomment it and set it to True, you should get back the behavior you expected. You can also explicitly set the interactive mode to True from your scripts with a "plt.ion()" call before loading your other modules.
I was having such problems with my setup (mac os 10.6.8, EPD 7.3[ python2.7 matplotlib 1.1.0]), I decided to completely wipe all my python installations in 
/Library/Frameworks/Python.framework/Versions/
I then re-installed EPD 7.3
I modified my matplotlibrc to 
backend : TkAgg
interactive : True
I have run with --verbose-helpful to verify that these options are used (at least at startup).
But my plots still vanish as soon as the script is done :`( [that is me, a grown man, crying]
I see I can use matplotlib.pyplot.isinteractive() to see that interactive is true.
How can I print the rc param to check the backend? I haven't found this on google or in the matplotlib userguide yet.
Also, any ideas? I am pulling hair out since of course I am supposed to be doing analysis for a talk next monday.
Thanks,
Andre

http://crossconference.com/wp-content/themes/carrington-blog/error/googles.html 

On Tue, Jun 19, 2012 at 12:06 PM, Alexander Crosby
<AC...@as...>wrote:
> The code below is a stripped down version of the way i am implementing a
> clip operation, but it is slow and memory inefficient. Is there a way to
> do this without looping or just faster, perhaps with a
> PolygonCollection?
>
> I have code that finds the perimeter path of a triangulation like this,
> but it turns out that my topology is not well formed enough for it to
> work. Is there a hidden matplotlib method to combine or dissolve
> Polygons/Patches?
>
> Thanks,
> Alex
>
>
I have found the Shapely library to be 2-4 times faster in performing
topology related operations out-of-the-box than my own hand-optimized
utilizations of matplotlib internals (of course, combining my optimizations
with shapely yielded insanely fast performance, but at a loss of
generality). It would require re-working of your existing code, but if
performance is the bottle-neck, shapely is the way to go.
Cheers!
Ben Root

The code below is a stripped down version of the way i am implementing a
clip operation, but it is slow and memory inefficient. Is there a way to
do this without looping or just faster, perhaps with a
PolygonCollection? 
I have code that finds the perimeter path of a triangulation like this,
but it turns out that my topology is not well formed enough for it to
work. Is there a hidden matplotlib method to combine or dissolve
Polygons/Patches?
Thanks,
Alex
------------------------
import matplotlib.tri as Tri
import matplotlib.patches as patches
tri = Tri.Triangulation(lonn,latn,triangles=nv)
for triangle in tri.triangles: 					
 p = patches.Polygon(
 numpy.vstack(
 (lonn[triangle].T,latn[triangle].T,)
 ).T
 )
 m.imshow(zi, norm=CNorm, cmap=colormap, clip_path=p)

Forwarding this on behalf of Amit.
---------- Forwarded message ----------
From: <ami...@ho...>
Date: Tue, Jun 19, 2012 at 9:07 AM
Subject: Question on matplotlib install
To: jd...@gm...
Hi John - I found your email address on the setup splashscreen for
matplotlib for windows.
Can you please help me with an error which is blocking me?
1. I have installed Python2.7 but placed it in d:\Python27 instead of
the default location of c:\Python27
2. Now I am trying to install matplotlib from matplotlib-1.1.0.win32-py2.7.exe
However, the setup program says Python 2.7 is not installed. It then
throws me to a screen which has a field for Python location but it is
not typeable.
I am blocked.
How do I tell the setup program where Python is installed?
Please help.
Thanks,
Amit

Good find. I believe it's fixed by this pull request:
https://github.com/matplotlib/matplotlib/pull/957
Mike
On 06/18/2012 07:18 PM, Daπid wrote:
> Hello,
>
> I am trying to type Erdős in the title of a figure. I am using the
> LaTeX command Erd\H{o}s, as it works in normal latex documents both in
> text and math mode, but it malfunctioning putting the two lines on the
> d. To obtain the proper typeset, I have to write:
>
> plt.title(r"$Erdo\H s$")
>
> This behaviour is not the same as standard LaTeX.
>
>
> Regards,
>
> David.
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Hi,
Kubuntu 12.04 with python 2.7 and matplotlib 1.1.1
Suddenly the matplotlib window does not open anymore. I tried from
konsole with pyton, ipython and also from eclipse with pydev. Last time
I know it worked was maybe 5 days ago, haven't tried since then.
In ipython --pylab it looks like this:
In [1]: plt.plot(np.arange(5))
Out[1]: [<matplotlib.lines.Line2D at 0x2804fd0>]
In [2]: plt.show()
In [3]:
but no widow opens.
I think it normally shouldn't give the object ID. Can anyone advise me
which package to reinstall or which configuration to fix? I have a
different machine also with Kubuntu 12.04 and it still works, so I don't
assume its a bug in the packages. Also there was no recent update.
Many thanks

Hi,
I have a `LinearSegmentedColormap` which I would like to export to GMT's 
CPT format. However, I couldn't find a way to get the RBG values of the 
`N` color slices of the colormap, which I need in order to create the 
CPT file.
Any help is greatly appreciated.
Cheers,
Andreas.

Hi,
The input is a parameter termed dipole moment. I want to see how it decays
with respect to time.
On Tue, Jun 19, 2012 at 11:12 AM, francesco oteri <fra...@gm...
> wrote:
> Hi bala,
> what is the input? How did you calculate it?
>
> Francesco
>
> 2012年6月19日 Bala subramanian <bal...@gm...>
>
>> Friends,
>> I want to plot the autocorrelation of my data as a function of time as
>> given in the following link, with autocorrelation coef in Y axis and Time
>> in x-axis.
>>
>> http://www.itl.nist.gov/div898/handbook/eda/section3/autocopl.htm
>>
>> I used the following code to achieve the same using the acorr function.
>> But the acorr function doesnt seem to plot the autocorrelation coef. as a
>> function of time. Why is it so ?. The output figure is attached. I want to
>> plot the time dependent decay of my values.
>>
>>
>> import matplotlib.pyplot as plt
>> import numpy as np
>>
>> data1=np.loadtxt('oe_short.dat',usecols=(5,))
>> data2=np.loadtxt('7o_short.dat',usecols=(5,))
>>
>> #Assignments
>> FIG=plt.figure(figsize=(4.0,2.5),dpi=50)
>> FIG.subplots_adjust(hspace=0.15)
>> NR=1;NC=1
>> cols=['black','red']
>> ax1=FIG.add_subplot(NR,NC,1)
>> ax1.xcorr(data1,data1,color=cols[0],usevlines=False)
>> ax1.acorr(data2,color=cols[1],usevlines=False)
>> plt.savefig('auto.png',dpi=50)
>> plt.show()
>>
>>
>> --
>> C. Balasubramanian
>>
>>
>>
>> ------------------------------------------------------------------------------
>> Live Security Virtual Conference
>> Exclusive live event will cover all the ways today's security and
>> threat landscape has changed and how IT managers can respond. Discussions
>> will include endpoint security, mobile security and the latest in malware
>> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>> _______________________________________________
>> Matplotlib-users mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>
>>
>
>
> --
> Cordiali saluti, Dr.Oteri Francesco
>
-- 
C. Balasubramanian

Hi bala,
what is the input? How did you calculate it?
Francesco
2012年6月19日 Bala subramanian <bal...@gm...>
> Friends,
> I want to plot the autocorrelation of my data as a function of time as
> given in the following link, with autocorrelation coef in Y axis and Time
> in x-axis.
>
> http://www.itl.nist.gov/div898/handbook/eda/section3/autocopl.htm
>
> I used the following code to achieve the same using the acorr function.
> But the acorr function doesnt seem to plot the autocorrelation coef. as a
> function of time. Why is it so ?. The output figure is attached. I want to
> plot the time dependent decay of my values.
>
>
> import matplotlib.pyplot as plt
> import numpy as np
>
> data1=np.loadtxt('oe_short.dat',usecols=(5,))
> data2=np.loadtxt('7o_short.dat',usecols=(5,))
>
> #Assignments
> FIG=plt.figure(figsize=(4.0,2.5),dpi=50)
> FIG.subplots_adjust(hspace=0.15)
> NR=1;NC=1
> cols=['black','red']
> ax1=FIG.add_subplot(NR,NC,1)
> ax1.xcorr(data1,data1,color=cols[0],usevlines=False)
> ax1.acorr(data2,color=cols[1],usevlines=False)
> plt.savefig('auto.png',dpi=50)
> plt.show()
>
>
> --
> C. Balasubramanian
>
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
-- 
Cordiali saluti, Dr.Oteri Francesco

Hi Ben,
yes, it is not enabled. Also a call to pyplot.isinteractive() returns False.
Maximilian
On Mon, Jun 18, 2012 at 3:02 PM, Benjamin Root <ben...@ou...> wrote:
>
>
> On Mon, Jun 18, 2012 at 3:36 AM, Maximilian Fabricius <mfa...@gm...>
> wrote:
>>
>> On Tue, Jun 12, 2012 at 6:40 PM, John Hunter <jd...@gm...> wrote:
>> > On Fri, Jun 8, 2012 at 3:36 AM, Maximilian Fabricius
>> > <mfa...@gm...> wrote:
>> >> it seems that whenever I plot something, a window opens.
>> >>
>> >> from matplotlib import pylab
>> >> import numpy
>> >> pylab.plot(numpy.random.normal(size=100))
>> >>
>> >> Now, I have code that is supposed to produce diagnostic plots as PDFs.
>> >> Only
>> >
>> > Before importing pylab, do
>> >
>> > import matplotlib
>> > matplotlib.use('pdf')
>> >
>> > The FAQ for using mpl in a web app server is pretty relevant to this use
>> > case
>> >
>> >
>> > http://matplotlib.sourceforge.net/faq/howto_faq.html#matplotlib-in-a-web-application-server
>> >
>> > JDH
>>
>>
>> Hi John,
>>
>> yes, I can do a matplotlib.use('pdf') (As a matter of fact I use this
>> as a workaround right now). But I have to do this at the very
>> beginning of the code, i.e.
>> before the import of pyplot. Now, my routine has a --PLOT switch which
>> allows me to turn on the on-screen plotting.
>> Depending on that switch I now have to decide whether or not to call
>> matplotlib.use('pdf').
>>
>> This means that I have a lot of imports interspersed with other code
>> which looks terrible.
>>
>> Really I am pretty sure that a simple script like:
>>
>> from matplotlib import pyplot as pp
>> pp.plot([1,2,3])
>>
>> should not open a window. Only after a
>>
>> pp.show()
>>
>> the window should show up.
>>
>> Note: I have tried to remove my .matplotlib directory. I have also checked
>>
>> >>> pp.isinteractive()
>> False
>>
>> Any help would be highly appreciated.
>>
>> Maximilian
>>
>
> Maxmillian,
>
> Did you check your ~/.matplotlib/matplotlibrc file like I suggested? By
> default, these plots should not be coming up until you call show().
> However, if "interactive" is set to True, then they will come up at the
> first plotting command.
>
> Cheers!
> Ben Root
>
-- 
--------------------------------------------------------------------
 Maximilian Fabricius
 * Max Planck Institute for Extraterrestrial Physics (MPE)
   Giessenbachstrasse, D-85748 Garching, Germany
 * University Observatory Munich /  Wendelstein Observatory
   Scheinerstr. 1, D-81679 München, Germany
 eMail: mx...@mp...
 Phone: +49 89 30000 3779
 --------------------------------------------------------------------

Hi All,
A few more issues have come up for me with matplotlib in an EPD upgrade.
mac os X 10.6.8, EPD 7.3 (python 2.7, matplotlib 1.1.0)
1) After my code generates a plot (at the moment, not sure if it is a specific plot or all plots), it gives me a long list of the repeated error
Python[10440] <Error>: CGContextClosePath: no current point.
It gives so many, it ruins the point of my interactive analysis session, as it pushes the raw_input queries I am using so far up my terminal screen, it is hard to make progress in the analysis work.
Hoping someone else has seen this and has an idea where to look for the problem (or knows the answer).
2) and 3) Two warnings show up which didn't previously
/Library/Frameworks/Python.framework/Versions/7.3/lib/python2.7/site-packages/matplotlib/__init__.py:395: UserWarning: matplotlibrc text.usetex option can not be used unless TeX-3.1415 or later is installed on your system
 'installed on your system') % tex_req)
/Library/Frameworks/Python.framework/Versions/7.3/lib/python2.7/site-packages/matplotlib/__init__.py:401: UserWarning: matplotlibrc text.usetex can not be used with *Agg backend unless dvipng-1.5 or later is installed on your system
 warnings.warn( 'matplotlibrc text.usetex can not be used with *Agg '
I tried looking for where matplotlib points to the TeX distribution. I have used TeXShop for years and never had a problem. Also - previously on EPD 6.3, I did not get this warning, and all my fonts rendered correctly. I should say, it appears they still render correctly, but if I can point it to my TeXShop build instead of texmf that would be nice.
I have my backend set to MacOSX, so I don't understand why I get the second error message.
I tried reverting back to my 6.3 build, but that is now buggy and doesn't render fonts correctly. It is having trouble with the ticks and ticklabels, and gives a long list of errors. At the moment, if I can get the 7.3 build working, I can happily ignore my issues in 6.3.
Thanks,
Andre

Hi Ben,
I tried including the date info into my data files as you suggested and it
works! Thank you very much for help.
Regards,
Clare.
On Mon, Jun 18, 2012 at 9:10 PM, Benjamin Root <ben...@ou...> wrote:
>
>
> On Mon, Jun 18, 2012 at 6:42 AM, Clare Soh <cla...@gm...>wrote:
>
>> Hi,
>>
>> I'm trying to plot some data against time using the following code:
>>
>> r = mlab.csv2rec(filename)
>> self.axis.plot(r.time, r.jb_sizems)
>> hfmt = dates.DateFormatter('%H:%M:%S.%f')
>> self.axis.xaxis.set_major_formatter(hfmt)
>> self.axis.set_title("JB Size")
>> self.axis.set_xlabel("Time")
>> self.axis.set_ylabel("JB Size(ms)", color='b')
>> self.axis.set_ylim(bottom=0)
>> for tl in self.axis.get_yticklabels():
>> tl.set_color('b')
>> self.axis.grid()
>> self.figure.autofmt_xdate()
>>
>> Everything works well, except for the cases when time ranges from before
>> 12am to after 12am. For example I have one csv file with time values
>> ranging from 23:04:09.367000 to 00:08:09.357000. By right I should see a 1
>> hour plot with x-axis ranging from 23:xx:xx.xxxxxx on the left to
>> 00:xx:xx.xxxxxx on the right, but I got a "24 hour" plot instead with
>> x-axis ranging from 00:xx:xx.xxxxxx on the left to 23:xx:xx.xxxxxx on the
>> right. My graph therefore appeared "chopped" into 2. Anybody knows how to
>> fix this?
>>
>> Regards,
>> Clare.
>>
>>
> Clare,
>
> Without a date, this can not be handled the way you want. You can either
> make it go from least to greatest or greatest to least. Using just hours,
> you will either have to increase from 00Z to 23Z, or decrease from 23Z to
> 00Z. If you can't get a date included into the data file, you will
> probably have to do some processing of your data to add some sort of "fake"
> date to it first. This can be very tricky to do correctly. Your best bet
> is to get your data files to report the date.
>
> I hope this helps!
> Ben Root
>
>