matplotlib-users Mailing List for matplotlib (Page 4)

  Hi all! I'm experiencing a strange behaviour with sankey diagram. As
you can see from the attached image, the patch label it's not
positioned in the middle of the patch (as it should), but it's plotted
over the first label... The code is attached also.
Any hints?

Thanks,
  Giovanni

Dear all,

Is there a way to remove colorbar? axes.cla() clears only the region for
map but not the colorbar.

Chao

-- 
***********************************************************************************
Chao YUE
Laboratoire des Sciences du Climat et de l'Environnement (LSCE-IPSL)
UMR 1572 CEA-CNRS-UVSQ
Batiment 712 - Pe 119
91191 GIF Sur YVETTE Cedex
Tel: (33) 01 69 08 29 02; Fax:01.69.08.77.16
************************************************************************************

On Mon, May 21, 2012 at 3:45 AM, Chao YUE <cha...@gm...> wrote:

> Dear all,
>
> I have a question on how to change default figure dpi when try to save
> figure in interactive mode.
> I try to use mat.rcParams['figure.dpi']=300 before I plot, then the
> problem is that the figure which is drawn is too big for interactive check
> (maybe because of increase in dpi).
> the mat.rcParams['figure.figsize'] is still default [8,6]. The other
> option is that always use fig=gcf(); and fig.savefig('my_plot.jpg',
> dpi=300).
> I am suing 'Agg' backend on ubuntu system.
>
> thanks et cheers,
>
> Chao
>
> Hi Chao,

Have you tried the 'savefig.dpi' rcParam? (see customization
docs<http://matplotlib.sourceforge.net/users/customizing.html>
)

Best,
-Tony

Dear all,

I have a question on how to change default figure dpi when try to save
figure in interactive mode.
I try to use mat.rcParams['figure.dpi']=300 before I plot, then the problem
is that the figure which is drawn is too big for interactive check (maybe
because of increase in dpi).
the mat.rcParams['figure.figsize'] is still default [8,6]. The other option
is that always use fig=gcf(); and fig.savefig('my_plot.jpg', dpi=300).
I am suing 'Agg' backend on ubuntu system.

thanks et cheers,

Chao
-- 
***********************************************************************************
Chao YUE
Laboratoire des Sciences du Climat et de l'Environnement (LSCE-IPSL)
UMR 1572 CEA-CNRS-UVSQ
Batiment 712 - Pe 119
91191 GIF Sur YVETTE Cedex
Tel: (33) 01 69 08 29 02; Fax:01.69.08.77.16
************************************************************************************

Hi Alejandro,

Thank you very much for your help! It is a font priority problem. I added
cmr to the first of font.serif. Problem solved!

rcParams['font.serif'] = ['Computer Modern Roman'] + rcParams['font.serif']

Many thanks again!

A Spherical Chicken

On Sun, May 20, 2012 at 6:55 PM, Alejandro Weinstein <
ale...@gm...> wrote:

> On Sun, May 20, 2012 at 4:14 PM, Yang Zhang
> <a.s...@gm...> wrote:
> > Recently I have tried to use latex to render all the texts in a plot for
> > consistent look with other texts in a paper. However, it seems that all
> the
> > texts in the plot are rendered in bold font. Please see the attached
> code.
>
> Can you try with the following setup?
>
> params = {'backend': 'Agg',
>         'ps.usedistiller' : 'xpdf',
>         'text.usetex' : True,
>         'font.family': 'serif',
>         'font.serif' : ['Times'],
>         }
> mpl.rcParams.update(params)
>
> I made all the figures in this paper using these parameters:
> http://arxiv.org/pdf/1110.5063v1
>
> Alejandro
>

On Sun, May 20, 2012 at 4:14 PM, Yang Zhang
<a.s...@gm...> wrote:
> Recently I have tried to use latex to render all the texts in a plot for
> consistent look with other texts in a paper. However, it seems that all the
> texts in the plot are rendered in bold font. Please see the attached code.

Can you try with the following setup?

params = {'backend': 'Agg',
         'ps.usedistiller' : 'xpdf',
         'text.usetex' : True,
         'font.family': 'serif',
         'font.serif' : ['Times'],
         }
mpl.rcParams.update(params)

I made all the figures in this paper using these parameters:
http://arxiv.org/pdf/1110.5063v1

Alejandro

Hi,

mpl is a great package. Thanks for the effort first!

Recently I have tried to use latex to render all the texts in a plot for
consistent look with other texts in a paper. However, it seems that all the
texts in the plot are rendered in bold font. Please see the attached code.

==
#!/opt/local/bin/python
from matplotlib.pyplot import *
from matplotlib import rc
rc('text', usetex=True)
rc('font', family='serif')

text(.1, .2, r'first $f(x,y)=x+y\ \mathrm{first}\ \mathbf{first}\
\textrm{first}\ \textnormal{first}$')
xlabel(r'first $f(x,y)=x+y\ \mathrm{first}\ \mathbf{first}\ \textrm{first}\
\textnormal{first}$')

savefig('test_tex.pdf', transparent=True)
==

The text in 1) normal environment (quoted in r'text'), 2) in \textrm{}, 3)
\textnormal{} are the same with 4) \mathbf{}. The only way to produce with
"normal" font is to put the text in \mathrm{}.

The matplotlibrc is default or blank. I was able to reproduce the result
with versions 1.1.0 and 1.1.1 on both linux and mac. And I couldn't find a
quick fix from the internet.

Furthermore, I found that the texts in the tutorial page
http://matplotlib.sourceforge.net/users/usetex.html are also bolded: "TeX
is Number ..." is thicker than the font in the following equation and the
tics labels, whose "normal" font can be seen by putting them in \mathrm{}.

It is quite annoying to have inconsistent font in the figure and the main
texts. Could someone look into it? Thanks a million!

A Spherical Chicken

Hello Michael,

On Thu, May 3, 2012 at 1:25 PM, Michael Droettboom <md...@st...> wrote:
> ipython - **seems unnecessary**

removed

> python-configobj - **necessary only for a long abandoned experimental
> version of matplotlib**

removed

> python-epydoc - **obsolete**

removed

> python-qt4 - **not needed for build**

without it, there's several errors building the docs when rendering
the backend_qt4agg page, so I'll leave it

> python-qt4-dev - **not needed for build**
> python-qt-dev - **obsolete**

removed

> python-traits (>= 2.0) - **not needed -- matplotlib doesn't use traits**

removed

> python-wxgtk2.8 - **not needed for build**

same as for

> python-wxgtk2.8-dbg - **not needed for build**

removed

Thanks for your review!

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

I was wondering whether this feature has been built-in via specgram (maybe
needs a transformation) or if somebody has wrote something to easily
implement. If not, the current best method of going about it as I don't see
it in the documentation.

Cheers,
Jeff

* Miro Ilias wrote:
> Hi,
> 
> I have the same problem on Windows 7: I installated Python 2.7 from
> installation package, and  matplotlib package gave me the mentioned
> error message during installation.
 
You could also try EPD, Enthought Python Distribution, which ships a
lot of modules in one package, i.e. numpy, scipy, sympy and
also matplotlib. You can download it here:

http://enthought.com/repo/free/

-- 
Answer:   Because we read from top to bottom, left to right!
Question: Why should i start my reply below the quoted text?

On Saturday, May 19, 2012, Miro Ilias wrote:

> Hi,
>
> I have the same problem on Windows 7: I installated Python 2.7 from
> installation
> package, and  matplotlib package gave me the mentioned error message during
> installation.
>
> Miro
>
>
>
Maybe this is a 32/64-bit issue? I.e., the python install was 64 bits but
the mpl installer was for 32 bits?

Ben Root

Hi,

I have the same problem on Windows 7: I installated Python 2.7 from installation
package, and  matplotlib package gave me the mentioned error message during
installation.

Miro

Just do something like this:

newx = concatenate((c_b211_jmk,c_b204_jmk))
newy = concatenate((c_b211_k, c_b204_k))
newc = concatenate((ones(c_b211_k.size),-1*ones_like(c_b204_k.size)))

hexbin(newx,newy,C=newc,reduce_C_function=np.sum)

Then the color coding in each bin should be the number of counts from
b211 - number of counts from b204.


On Fri, May 18, 2012 at 12:11 PM, Sebastian <se...@gm...> wrote:
> Any hints on how to visualize the density difference between two hexbin
> plot, each with x-y (2D) data?
>
> Each set has roughly 300,000 points.
> The range in x and y values for each data set are roughly similar but with
> slightly different density:
> range x,x1: -1.222 to 3.656
> range y,y1: 13.191, 18.150
>
> So the steps:
>
> (1) Produce the two hexbin maps:
>
> fig1=hexbin(c_b204_jmk,c_b204_k,C = None, gridsize = 100, bins = None,
> xscale = 'linear', yscale = 'linear', cmap=None, norm=None, vmin=None,
> vmax=None, alpha=None, linewidths=None, edgecolors='none',reduce_C_function
> = np.mean, mincnt=None, marginals=False)
>
> fig2=hexbin(c_b211_jmk,c_b211_k,C = None, gridsize = 100, bins = None,
> xscale = 'linear', yscale = 'linear', cmap=None, norm=None, vmin=None,
> vmax=None, alpha=None, linewidths=None, edgecolors='none',reduce_C_function
> = np.mean, mincnt=None, marginals=False)
>
> (2) Determine the difference in the hexbin counts, with
>
> diff_hex=fig2.get_array()-fig1.get_array()
>
>
> (3) BUT when i try to plot the diff hexbin map
>
> fig3=hexbin(c_b211_jmk,c_b211_k,C = diff_hex, gridsize = 100, bins = None,
> xscale = 'linear', yscale = 'linear', cmap=None, norm=None, vmin=None,
> vmax=None, alpha=None, linewidths=None, edgecolors='none',reduce_C_function
> = np.mean, mincnt=None, marginals=False)
>
> I obviously get a:
> IndexError: index out of bounds
>
> Because:
> len(c_b211_jmk) = len(c_b211_k)  != len(diff_hex),
> Since diff_hex are the binned counts.
>
> (4) Any simple way to get around this, to plot the hexbin difference counts
> on top of the the  hexbin (c_b211_jmk,c_b211_k) distribution?
>
> thanks in advance & with best regards,
> - Sebastian



-- 
Erik Tollerud

Available for download at 
https://sourceforge.net/projects/matplotlib/files/matplotlib-toolkits/basemap-1.0.3/

Thanks to Christoph Gohlke for making windows installers.

Highlights:
   - new 'round' keyword for polar-centric plots (see polarmaps.py example).
   - the coastline filling bug has now finally been (mostly) fixed 
(there are still probably
     some corner cases that trigger it).
   - coastline and political boundary databases updated.
   - examples now work in python 3
   - added hexbin and streamplot methods.
   - added kav7 (Kavrayskiy VII) and eck4 (Eckert IV) projections.
     - UTM zones can now be plotted (see utmtest.py example).
     - many bug fixes.


full Changelog:

version 1.0.3 (git tag v1.0.3rel)
--------------------------------
* fix some more python 3 compatibility issues (all examples now
   work with python 3.2).
* added alpha keyword to fillcontinents (to set transparency).
* update geos from 3.3.1 to 3.3.3.
* upgrade proj4 source to version 4.8.0, pyproj to version 1.9.2.
* add k_0 keyword for projection = 'tmerc' so UTM zones can be
   created. Also can be used with 'lcc','omerc' and 'stere'.
   Added "utmtest.py" example.
* add streamplot method, along with streamplot_demo.py example.
* add boundarylats, boundarylons Basemap attributes (arrays
   describing map boundaries - useful for illustrating map projection
   region on another map). Example illustrating usage
   (make_inset.py) added.
* update coastlines, rivers, political boundaries to GSHHS
   2.2.0/GMT 4.5.7. The fillcontinents bug (filling the outside
   instead of the inside of a coastline polygon) is now much
   harder to trigger.
* add 'round' keyword keyword to Basemap.__init__ for pole-centered
   projections to make them round (clipped at boundinglat) instead
   of square.
* added hexbin method, along with hexbin_demo.py example.
* drawmapboundary now uses axes bgcolor as default fill_color. If
   no color fill wanted, set fill_color='none' (a string).
* clip coastlines for nplaea,npaeqd,splaea,spaeqd in stereographic
   coordinates to avoid S. America disappearing in some south polar
   plots.
* fix broken daynight terminator function.
* added kav7 (Kavrayskiy VII) and eck4 (Eckert IV) projections
   (https://github.com/matplotlib/basemap/issues/9).
* update pyproj source from pyproj.googlecode.com. Includes
   new more robust and accurate pure python code for geodesic computations
   from geographiclib.
* bug fixes for celestial projections, and non-standard sphere
   radii (https://github.com/matplotlib/basemap/pull/6).
* fix bug in drawparallels that results in 'KeyError' when drawing
   parallels very close together (0.1 degrees).
* fix constant in Robinson projection (update PJ_robin.c
   from proj4 svn).
* fix typo in setup.py (replace "!= ['sdist','clean']" with
   "not in ['sdist','clean']").
* make sure drawmeridians can handle wrap-around (so that
   if projection is defined in -180 to 0 and user asks for
   meridians from 180 to 360 to be drawn, it should work).
   Only affects projections 'mill','gall','merc' and 'cyl'.

Regards,

Jeff

-- 
Jeffrey S. Whitaker         Phone  : (303)497-6313
Meteorologist               FAX    : (303)497-6449
NOAA/OAR/PSD  R/PSD1        Email  : Jef...@no...
325 Broadway                Office : Skaggs Research Cntr 1D-113
Boulder, CO, USA 80303-3328 Web    : http://tinyurl.com/5telg

Any hints on how to visualize the density difference between two hexbin
plot, each with x-y (2D) data?

Each set has roughly 300,000 points.
The range in x and y values for each data set are roughly similar but with
slightly different density:
range x,x1: -1.222 to 3.656
range y,y1: 13.191, 18.150

So the steps:

(1) Produce the two hexbin maps:

fig1=hexbin(c_b204_jmk,c_b204_k,C = None, gridsize = 100, bins = None,
xscale = 'linear', yscale = 'linear', cmap=None, norm=None, vmin=None,
vmax=None, alpha=None, linewidths=None, edgecolors='none',reduce_C_function
= np.mean, mincnt=None, marginals=False)

fig2=hexbin(c_b211_jmk,c_b211_k,C = None, gridsize = 100, bins = None,
xscale = 'linear', yscale = 'linear', cmap=None, norm=None, vmin=None,
vmax=None, alpha=None, linewidths=None, edgecolors='none',reduce_C_function
= np.mean, mincnt=None, marginals=False)

(2) Determine the difference in the hexbin counts, with

diff_hex=fig2.get_array()-fig1.get_array()


(3) BUT when i try to plot the diff hexbin map

fig3=hexbin(c_b211_jmk,c_b211_k,C = diff_hex, gridsize = 100, bins = None,
xscale = 'linear', yscale = 'linear', cmap=None, norm=None, vmin=None,
vmax=None, alpha=None, linewidths=None, edgecolors='none',reduce_C_function
= np.mean, mincnt=None, marginals=False)

I obviously get a:
IndexError: index out of bounds

Because:
len(c_b211_jmk) = len(c_b211_k)  != len(diff_hex),
Since diff_hex are the binned counts.

(4) Any simple way to get around this, to plot the hexbin difference counts
on top of the the  hexbin (c_b211_jmk,c_b211_k) distribution?

thanks in advance & with best regards,
- Sebastian

I think having only the one failure (and one from a test that is quite 
new) is pretty good.

I'm  not sure why that test is failing -- is it possible it's testing 
with an earlier checkout using tests from a newer checkout.  This would 
happen, for example, if running the tests from the matplotlib source 
directory.

Mike

On 05/15/2012 12:08 PM, Edward C. Jones wrote:
> I use up-to-date Debian testing (wheezy) with the amd64 architecture
> and Debian's python3.2.  I install matplotlib from the tarball
>       matplotlib-matplotlib-v1.1.0-684-ge87374e.tar.gz
> Before the current install, I had also on my system Debian's
> python-matplotlib and python-matplotlib-data packages.  They contain
> some of the same
> files as the tarball (/etc/matplotlibrc).  I have completely removed the two
> Debian packages. My Debian system includes the packages python3-tk, tck8.5,
> tcl8.5-dev, tk8.5, and tk-dev.
>
> I did a new matplotlib install, starting by unpacking the tarball.
>
> In /usr/local/lib/python3.2/dist-packages/matplotlib/tests/test_text.py
> in test_afm_kerning, I added before the assert:
>       xxx = afm.string_width_height('VAVAVAVAVAVA')
>       print(xxx)
>
> The results are:
>
>   >  python3.2
> Python 3.2.3rc2 (default, Mar 21 2012, 05:47:04)
> [GCC 4.6.3] on linux2
> Type "help", "copyright", "credits" or "license" for more information.
>   >>>  import matplotlib
>   >>>  matplotlib.test()
> ..K........./usr/lib/python3/dist-packages/nose/tools.py:82:
> ResourceWarning: unclosed file<_io.BufferedRandom name=3>
>     pass
> ...../usr/lib/python3.2/subprocess.py:650: ResourceWarning: unclosed file
> <_io.FileIO name=6 mode='rb'>
>     _cleanup()
> /usr/lib/python3.2/subprocess.py:650: ResourceWarning: unclosed file
> <_io.FileIO name=8 mode='rb'>
>     _cleanup()
>
> Many "."s and "K"s are printed.
>
> ======================================================================
> FAIL: matplotlib.tests.test_text.test_afm_kerning
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>     File "/usr/lib/python3/dist-packages/nose/case.py", line 198, in runTest
>       self.test(*self.arg)
>     File
> "/usr/local/lib/python3.2/dist-packages/matplotlib/tests/test_text.py", line
> 109, in test_afm_kerning
>       assert afm.string_width_height('VAVAVAVAVAVA') == (7174.0, 718)
> AssertionError: AssertionError:
> -------------------->>  begin captured stdout<<  ---------------------
> (8004.0, 718)
>
> --------------------->>  end captured stdout<<  ----------------------
>
> ----------------------------------------------------------------------
> Ran 1091 tests in 312.866s
>
> FAILED (KNOWNFAIL=275, failures=1)
> /usr/local/lib/python3.2/dist-packages/matplotlib/__init__.py:937:
> /UserWarning:  This call to matplotlib.use() has no effect
> because the the backend has already been chosen;
> matplotlib.use() must be called *before* pylab, matplotlib.pyplot,
> or matplotlib.backends is imported for the first time.
>
>     if warn: warnings.warn(_use_error_msg)
> False
>   >>>
> =======
> Here is one of matplotlib's simple sample programs.  It works.
>
> #! /usr/bin/env python3.2
>
> from pylab import *
>
> t = arange(0.0, 2.0, 0.01)
> s = sin(2*pi*t)
> plot(t, s, linewidth=1.0)
>
> xlabel('time (s)')
> ylabel('voltage (mV)')
> title('About as simple as it gets, folks')
> grid(True)
> show()
>
>
> ------------------------------------------------------------------------------
> 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

On Wed, May 16, 2012 at 9:30 AM, Michael Droettboom <md...@st...> wrote:
>
>
> Or, in an existing clone of the main repository, add my fork as a remote
>
>     git remote add mdboom git://github.com/mdboom/matplotlib.git
>     git fetch mdboom
>     git checkout mdboom/clipping-bug
>

Here are my steps following your 2nd suggestion:

1-) Cloned the master:
git clone git://github.com/matplotlib/matplotlib.git

2-) go into matplotlib dir and then execute:
sudo python setupegg.py develop

Tested my existing code and verified that the plotting error I reported in
the first message was still there.

3-) in the matplotlib dir I executed the 3 commands you typed to get your
fork

4-) Removed the build dir in matplotlib folder then re-executed setupegg.py
script

5-) Testing with your change my plot looks fine now, lines are drawn
correctly.

Thanks for easy to follow instructions and quick response.

On 05/16/2012 11:15 AM, Gökhan Sever wrote:
> Hmm, how can I test this change the easiest way?
>
> Clone the master and replace with your changes? or can I directly 
> clone your experimental branch?

You can either clone my fork and then checkout the branch with the change:

     git checkout clipping-bug

Or, in an existing clone of the main repository, add my fork as a remote

     git remote add mdboom git://github.com/mdboom/matplotlib.git
     git fetch mdboom
     git checkout mdboom/clipping-bug

Or, since the diff is only a few lines in path_converters.h, you could 
just apply it manually.

Be sure to remove your build directory before rebuilding: distutils 
doesn't pick up header file changes.

Mike

>
>
> On Wed, May 16, 2012 at 8:52 AM, Michael Droettboom <md...@st... 
> <mailto:md...@st...>> wrote:
>
>     I have a proposed solution here:
>
>     https://github.com/matplotlib/matplotlib/pull/872
>
>     Git bisect found that the first commit where this happens was here:
>
>     https://github.com/matplotlib/matplotlib/commit/4cd75cdf
>
>     This is the script I used to reproduce -- I assume it's the same
>     thing you're seeing:
>
>     from matplotlib import pyplot as plt
>     import numpy as np
>
>     x = np.linspace(0, 3.14 * 2, 3000)
>     y = np.sin(x)
>     x[::100] = np.nan
>     plt.plot(x, y)
>     plt.ylim(-0.25, 0.25)
>     plt.show()
>
>     Mike
>
>
>     On 05/16/2012 10:44 AM, Gökhan Sever wrote:
>>     Hi Mike,
>>
>>     Could you inform me about your progress? I can test your sample
>>     script. I was thinking to test from v1.1.x branch downwards to
>>     spot the source of the issue, but I just don't know how to clone
>>     at particular commit in git.
>>
>>     Thank you.
>>
>>     On Wed, May 16, 2012 at 6:51 AM, Michael Droettboom
>>     <md...@st... <mailto:md...@st...>> wrote:
>>
>>         Nevermind -- I've got something to reproduce this and am
>>         looking into it now.
>>
>>         Mike
>>
>>
>>         On 05/16/2012 08:13 AM, Michael Droettboom wrote:
>>>         On 05/15/2012 07:57 PM, Gökhan Sever wrote:
>>>>         Hello,
>>>>
>>>>         I have encountered a weird plotting issue recently using a
>>>>         recent mpl clone. See the linked pdfs for better
>>>>         demonstration of the issue:
>>>>
>>>>         http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_newmpl.pdf
>>>>         <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_newmpl.pdf>
>>>>         http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_oldmpl.pdf
>>>>         <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_oldmpl.pdf>
>>>>
>>>>
>>>>         newmpl file is created using the latest master branch
>>>>         (cloned and setup today)
>>>>         oldmpl is created using mpl v1.1.0
>>>>         (https://github.com/downloads/matplotlib/matplotlib/matplotlib-1.1.0.tar.gz)
>>>>
>>>>         Scroll down to page 4 in each file and you will see the
>>>>         wrong plotted behavior of alwp_lcl (black line) variable on
>>>>         newmpl file comparing to the correct version that is shown
>>>>         on oldmpl.
>>>>
>>>>         I was trying to figure out a way to correct this and I
>>>>         raised y-axis max to 2400 and then the line looks fine.
>>>>         However I have other data that show similar
>>>>         wrong behaviors, so I decided to try earlier mpl versions
>>>>         since I know that those plots were looking correct earlier
>>>>         (at least a few months back). Trying v1.1.x branch gave me
>>>>         the same results. Note that these data contain "nans". Are
>>>>         nan handling changed in recent mpl code or the way the data
>>>>         is plotted out of margins? I can't reproduce this
>>>>         with synthetic data.
>>>>
>>>         There have been changes to that code lately.  Is there any
>>>         way you can pack up a small script and data to reproduce
>>>         this?  Then I can poke at it and see what I find (it would
>>>         also make a good regression test).
>>>
>>>         Mike
>>>
>>>
>>>         ------------------------------------------------------------------------------
>>>         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...  <mailto:Mat...@li...>
>>>         https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>
>>
>>         ------------------------------------------------------------------------------
>>         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...
>>         <mailto:Mat...@li...>
>>         https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>
>>
>>
>>
>>     -- 
>>     Gökhan
>
>
>
>
> -- 
> Gökhan

Bisecting is definitely a better idea than my one-by-one setup iteration :)

Thanks for sharing the tip.

On Wed, May 16, 2012 at 8:54 AM, Michael Droettboom <md...@st...> wrote:

>  On 05/16/2012 10:44 AM, Gökhan Sever wrote:
>
> Could you inform me about your progress? I can test your sample script. I
> was thinking to test from v1.1.x branch downwards to spot the source of the
> issue, but I just don't know how to clone at particular commit in git.
>
>
> Also, to answer this question directly -- "git bisect" is a great way to
> find this:
>
> http://git-scm.com/book/en/Git-Tools-Debugging-with-Git#Binary-Search
>
> Cheers,
> Mike
>
>
>
>  Thank you.
>
> On Wed, May 16, 2012 at 6:51 AM, Michael Droettboom <md...@st...>wrote:
>
>>  Nevermind -- I've got something to reproduce this and am looking into it
>> now.
>>
>> Mike
>>
>>
>> On 05/16/2012 08:13 AM, Michael Droettboom wrote:
>>
>>  On 05/15/2012 07:57 PM, Gökhan Sever wrote:
>>
>> Hello,
>>
>>  I have encountered a weird plotting issue recently using a recent mpl
>> clone. See the linked pdfs for better demonstration of the issue:
>>
>>  http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_newmpl.pdf
>> http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_oldmpl.pdf
>>
>>
>>  newmpl file is created using the latest master branch (cloned and setup
>> today)
>> oldmpl is created using mpl v1.1.0 (
>> https://github.com/downloads/matplotlib/matplotlib/matplotlib-1.1.0.tar.gz
>> )
>>
>>  Scroll down to page 4 in each file and you will see the wrong
>> plotted behavior of alwp_lcl (black line) variable on newmpl file comparing
>> to the correct version that is shown on oldmpl.
>>
>>  I was trying to figure out a way to correct this and I raised y-axis
>> max to 2400 and then the line looks fine. However I have other data that
>> show similar wrong behaviors, so I decided to try earlier mpl versions
>> since I know that those plots were looking correct earlier (at least a few
>> months back). Trying v1.1.x branch gave me the same results. Note that
>> these data contain "nans". Are nan handling changed in recent mpl code or
>> the way the data is plotted out of margins? I can't reproduce this
>> with synthetic data.
>>
>>   There have been changes to that code lately.  Is there any way you can
>> pack up a small script and data to reproduce this?  Then I can poke at it
>> and see what I find (it would also make a good regression test).
>>
>> Mike
>>
>>
>>  ------------------------------------------------------------------------------
>> 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 lis...@li...https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>
>>
>>
>>
>> ------------------------------------------------------------------------------
>> 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
>>
>>
>
>
>  --
> Gökhan
>
>
>


-- 
Gökhan

Hmm, how can I test this change the easiest way?

Clone the master and replace with your changes? or can I directly clone
your experimental branch?


On Wed, May 16, 2012 at 8:52 AM, Michael Droettboom <md...@st...> wrote:

>  I have a proposed solution here:
>
> https://github.com/matplotlib/matplotlib/pull/872
>
> Git bisect found that the first commit where this happens was here:
>
> https://github.com/matplotlib/matplotlib/commit/4cd75cdf
>
> This is the script I used to reproduce -- I assume it's the same thing
> you're seeing:
>
> from matplotlib import pyplot as plt
> import numpy as np
>
> x = np.linspace(0, 3.14 * 2, 3000)
> y = np.sin(x)
> x[::100] = np.nan
> plt.plot(x, y)
> plt.ylim(-0.25, 0.25)
> plt.show()
>
> Mike
>
>
> On 05/16/2012 10:44 AM, Gökhan Sever wrote:
>
> Hi Mike,
>
>  Could you inform me about your progress? I can test your sample script.
> I was thinking to test from v1.1.x branch downwards to spot the source of
> the issue, but I just don't know how to clone at particular commit in git.
>
>  Thank you.
>
> On Wed, May 16, 2012 at 6:51 AM, Michael Droettboom <md...@st...>wrote:
>
>>  Nevermind -- I've got something to reproduce this and am looking into it
>> now.
>>
>> Mike
>>
>>
>> On 05/16/2012 08:13 AM, Michael Droettboom wrote:
>>
>>  On 05/15/2012 07:57 PM, Gökhan Sever wrote:
>>
>> Hello,
>>
>>  I have encountered a weird plotting issue recently using a recent mpl
>> clone. See the linked pdfs for better demonstration of the issue:
>>
>>  http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_newmpl.pdf
>> http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_oldmpl.pdf
>>
>>
>>  newmpl file is created using the latest master branch (cloned and setup
>> today)
>> oldmpl is created using mpl v1.1.0 (
>> https://github.com/downloads/matplotlib/matplotlib/matplotlib-1.1.0.tar.gz
>> )
>>
>>  Scroll down to page 4 in each file and you will see the wrong
>> plotted behavior of alwp_lcl (black line) variable on newmpl file comparing
>> to the correct version that is shown on oldmpl.
>>
>>  I was trying to figure out a way to correct this and I raised y-axis
>> max to 2400 and then the line looks fine. However I have other data that
>> show similar wrong behaviors, so I decided to try earlier mpl versions
>> since I know that those plots were looking correct earlier (at least a few
>> months back). Trying v1.1.x branch gave me the same results. Note that
>> these data contain "nans". Are nan handling changed in recent mpl code or
>> the way the data is plotted out of margins? I can't reproduce this
>> with synthetic data.
>>
>>   There have been changes to that code lately.  Is there any way you can
>> pack up a small script and data to reproduce this?  Then I can poke at it
>> and see what I find (it would also make a good regression test).
>>
>> Mike
>>
>>
>>  ------------------------------------------------------------------------------
>> 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 lis...@li...https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>
>>
>>
>>
>> ------------------------------------------------------------------------------
>> 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
>>
>>
>
>
>  --
> Gökhan
>
>
>


-- 
Gökhan

On 05/16/2012 10:44 AM, Gökhan Sever wrote:
> Could you inform me about your progress? I can test your sample 
> script. I was thinking to test from v1.1.x branch downwards to spot 
> the source of the issue, but I just don't know how to clone at 
> particular commit in git.

Also, to answer this question directly -- "git bisect" is a great way to 
find this:

http://git-scm.com/book/en/Git-Tools-Debugging-with-Git#Binary-Search

Cheers,
Mike

>
> Thank you.
>
> On Wed, May 16, 2012 at 6:51 AM, Michael Droettboom <md...@st... 
> <mailto:md...@st...>> wrote:
>
>     Nevermind -- I've got something to reproduce this and am looking
>     into it now.
>
>     Mike
>
>
>     On 05/16/2012 08:13 AM, Michael Droettboom wrote:
>>     On 05/15/2012 07:57 PM, Gökhan Sever wrote:
>>>     Hello,
>>>
>>>     I have encountered a weird plotting issue recently using a
>>>     recent mpl clone. See the linked pdfs for better demonstration
>>>     of the issue:
>>>
>>>     http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_newmpl.pdf
>>>     <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_newmpl.pdf>
>>>     http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_oldmpl.pdf
>>>     <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_oldmpl.pdf>
>>>
>>>
>>>     newmpl file is created using the latest master branch (cloned
>>>     and setup today)
>>>     oldmpl is created using mpl v1.1.0
>>>     (https://github.com/downloads/matplotlib/matplotlib/matplotlib-1.1.0.tar.gz)
>>>
>>>     Scroll down to page 4 in each file and you will see the wrong
>>>     plotted behavior of alwp_lcl (black line) variable on newmpl
>>>     file comparing to the correct version that is shown on oldmpl.
>>>
>>>     I was trying to figure out a way to correct this and I raised
>>>     y-axis max to 2400 and then the line looks fine. However I have
>>>     other data that show similar wrong behaviors, so I decided to
>>>     try earlier mpl versions since I know that those plots were
>>>     looking correct earlier (at least a few months back). Trying
>>>     v1.1.x branch gave me the same results. Note that these data
>>>     contain "nans". Are nan handling changed in recent mpl code or
>>>     the way the data is plotted out of margins? I can't reproduce
>>>     this with synthetic data.
>>>
>>     There have been changes to that code lately.  Is there any way
>>     you can pack up a small script and data to reproduce this?  Then
>>     I can poke at it and see what I find (it would also make a good
>>     regression test).
>>
>>     Mike
>>
>>
>>     ------------------------------------------------------------------------------
>>     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...  <mailto:Mat...@li...>
>>     https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>     ------------------------------------------------------------------------------
>     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...
>     <mailto:Mat...@li...>
>     https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>
>
> -- 
> Gökhan

I have a proposed solution here:

https://github.com/matplotlib/matplotlib/pull/872

Git bisect found that the first commit where this happens was here:

https://github.com/matplotlib/matplotlib/commit/4cd75cdf

This is the script I used to reproduce -- I assume it's the same thing 
you're seeing:

from matplotlib import pyplot as plt
import numpy as np

x = np.linspace(0, 3.14 * 2, 3000)
y = np.sin(x)
x[::100] = np.nan
plt.plot(x, y)
plt.ylim(-0.25, 0.25)
plt.show()

Mike

On 05/16/2012 10:44 AM, Gökhan Sever wrote:
> Hi Mike,
>
> Could you inform me about your progress? I can test your sample 
> script. I was thinking to test from v1.1.x branch downwards to spot 
> the source of the issue, but I just don't know how to clone at 
> particular commit in git.
>
> Thank you.
>
> On Wed, May 16, 2012 at 6:51 AM, Michael Droettboom <md...@st... 
> <mailto:md...@st...>> wrote:
>
>     Nevermind -- I've got something to reproduce this and am looking
>     into it now.
>
>     Mike
>
>
>     On 05/16/2012 08:13 AM, Michael Droettboom wrote:
>>     On 05/15/2012 07:57 PM, Gökhan Sever wrote:
>>>     Hello,
>>>
>>>     I have encountered a weird plotting issue recently using a
>>>     recent mpl clone. See the linked pdfs for better demonstration
>>>     of the issue:
>>>
>>>     http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_newmpl.pdf
>>>     <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_newmpl.pdf>
>>>     http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_oldmpl.pdf
>>>     <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_oldmpl.pdf>
>>>
>>>
>>>     newmpl file is created using the latest master branch (cloned
>>>     and setup today)
>>>     oldmpl is created using mpl v1.1.0
>>>     (https://github.com/downloads/matplotlib/matplotlib/matplotlib-1.1.0.tar.gz)
>>>
>>>     Scroll down to page 4 in each file and you will see the wrong
>>>     plotted behavior of alwp_lcl (black line) variable on newmpl
>>>     file comparing to the correct version that is shown on oldmpl.
>>>
>>>     I was trying to figure out a way to correct this and I raised
>>>     y-axis max to 2400 and then the line looks fine. However I have
>>>     other data that show similar wrong behaviors, so I decided to
>>>     try earlier mpl versions since I know that those plots were
>>>     looking correct earlier (at least a few months back). Trying
>>>     v1.1.x branch gave me the same results. Note that these data
>>>     contain "nans". Are nan handling changed in recent mpl code or
>>>     the way the data is plotted out of margins? I can't reproduce
>>>     this with synthetic data.
>>>
>>     There have been changes to that code lately.  Is there any way
>>     you can pack up a small script and data to reproduce this?  Then
>>     I can poke at it and see what I find (it would also make a good
>>     regression test).
>>
>>     Mike
>>
>>
>>     ------------------------------------------------------------------------------
>>     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...  <mailto:Mat...@li...>
>>     https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>     ------------------------------------------------------------------------------
>     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...
>     <mailto:Mat...@li...>
>     https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>
>
> -- 
> Gökhan

Hi Mike,

Could you inform me about your progress? I can test your sample script. I
was thinking to test from v1.1.x branch downwards to spot the source of the
issue, but I just don't know how to clone at particular commit in git.

Thank you.

On Wed, May 16, 2012 at 6:51 AM, Michael Droettboom <md...@st...> wrote:

>  Nevermind -- I've got something to reproduce this and am looking into it
> now.
>
> Mike
>
>
> On 05/16/2012 08:13 AM, Michael Droettboom wrote:
>
> On 05/15/2012 07:57 PM, Gökhan Sever wrote:
>
> Hello,
>
>  I have encountered a weird plotting issue recently using a recent mpl
> clone. See the linked pdfs for better demonstration of the issue:
>
>  http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_newmpl.pdf
> http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_oldmpl.pdf
>
>
>  newmpl file is created using the latest master branch (cloned and setup
> today)
> oldmpl is created using mpl v1.1.0 (
> https://github.com/downloads/matplotlib/matplotlib/matplotlib-1.1.0.tar.gz
> )
>
>  Scroll down to page 4 in each file and you will see the wrong
> plotted behavior of alwp_lcl (black line) variable on newmpl file comparing
> to the correct version that is shown on oldmpl.
>
>  I was trying to figure out a way to correct this and I raised y-axis max
> to 2400 and then the line looks fine. However I have other data that show
> similar wrong behaviors, so I decided to try earlier mpl versions since I
> know that those plots were looking correct earlier (at least a few months
> back). Trying v1.1.x branch gave me the same results. Note that these data
> contain "nans". Are nan handling changed in recent mpl code or the way the
> data is plotted out of margins? I can't reproduce this with synthetic data.
>
>   There have been changes to that code lately.  Is there any way you can
> pack up a small script and data to reproduce this?  Then I can poke at it
> and see what I find (it would also make a good regression test).
>
> Mike
>
>
> ------------------------------------------------------------------------------
> 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 lis...@li...https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>
>
> ------------------------------------------------------------------------------
> 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
>
>


-- 
Gökhan

Nevermind -- I've got something to reproduce this and am looking into it 
now.

Mike

On 05/16/2012 08:13 AM, Michael Droettboom wrote:
> On 05/15/2012 07:57 PM, Gökhan Sever wrote:
>> Hello,
>>
>> I have encountered a weird plotting issue recently using a recent mpl 
>> clone. See the linked pdfs for better demonstration of the issue:
>>
>> http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_newmpl.pdf 
>> <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_newmpl.pdf>
>> http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_oldmpl.pdf 
>> <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_oldmpl.pdf>
>>
>>
>> newmpl file is created using the latest master branch (cloned and 
>> setup today)
>> oldmpl is created using mpl v1.1.0 
>> (https://github.com/downloads/matplotlib/matplotlib/matplotlib-1.1.0.tar.gz)
>>
>> Scroll down to page 4 in each file and you will see the wrong 
>> plotted behavior of alwp_lcl (black line) variable on newmpl file 
>> comparing to the correct version that is shown on oldmpl.
>>
>> I was trying to figure out a way to correct this and I raised y-axis 
>> max to 2400 and then the line looks fine. However I have other data 
>> that show similar wrong behaviors, so I decided to try earlier mpl 
>> versions since I know that those plots were looking correct earlier 
>> (at least a few months back). Trying v1.1.x branch gave me the same 
>> results. Note that these data contain "nans". Are nan handling 
>> changed in recent mpl code or the way the data is plotted out of 
>> margins? I can't reproduce this with synthetic data.
>>
> There have been changes to that code lately.  Is there any way you can 
> pack up a small script and data to reproduce this?  Then I can poke at 
> it and see what I find (it would also make a good regression test).
>
> Mike
>
>
> ------------------------------------------------------------------------------
> 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

On 05/15/2012 07:57 PM, Gökhan Sever wrote:
> Hello,
>
> I have encountered a weird plotting issue recently using a recent mpl 
> clone. See the linked pdfs for better demonstration of the issue:
>
> http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_newmpl.pdf 
> <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_newmpl.pdf>
> http://atmos.uwyo.edu/~gsever/data/vocals_RF04_NU05_oldmpl.pdf 
> <http://atmos.uwyo.edu/%7Egsever/data/vocals_RF04_NU05_oldmpl.pdf>
>
>
> newmpl file is created using the latest master branch (cloned and 
> setup today)
> oldmpl is created using mpl v1.1.0 
> (https://github.com/downloads/matplotlib/matplotlib/matplotlib-1.1.0.tar.gz)
>
> Scroll down to page 4 in each file and you will see the wrong 
> plotted behavior of alwp_lcl (black line) variable on newmpl file 
> comparing to the correct version that is shown on oldmpl.
>
> I was trying to figure out a way to correct this and I raised y-axis 
> max to 2400 and then the line looks fine. However I have other data 
> that show similar wrong behaviors, so I decided to try earlier mpl 
> versions since I know that those plots were looking correct earlier 
> (at least a few months back). Trying v1.1.x branch gave me the same 
> results. Note that these data contain "nans". Are nan handling changed 
> in recent mpl code or the way the data is plotted out of margins? I 
> can't reproduce this with synthetic data.
>
There have been changes to that code lately.  Is there any way you can 
pack up a small script and data to reproduce this?  Then I can poke at 
it and see what I find (it would also make a good regression test).

Mike