matplotlib-users Mailing List for matplotlib

Could someone please kindly let me know if matplotlib is compatible with
Cygwin?  I installed all the necessary dependencies, yet I cannot get
matplotlib to compile correctly.  At the build step, this is what I get:

$ python setup.py build
basedirlist is: ['/usr/local', '/usr']
============================================================================
BUILDING MATPLOTLIB
            matplotlib: 1.0.0
                python: 2.6.5 (r265:79063, Jun 12 2010, 17:07:01)  [GCC
                        4.3.4 20090804 (release) 1]
              platform: cygwin

REQUIRED DEPENDENCIES
                 numpy: 1.4.1
             freetype2: found, but unknown version (no pkg-config)
                        * WARNING: Could not find 'freetype2' headers in any
                        * of '/usr/include', '.', '/usr/include/freetype2',
                        * './freetype2'.

OPTIONAL BACKEND DEPENDENCIES
                libpng: found, but unknown version (no pkg-config)
                        * Could not find 'libpng' headers in any of
                        * '/usr/include', '.'
cygwin warning:
  MS-DOS style path detected: C:/Cygwin/usr/share/tcl8.4/tclConfig.sh
  Preferred POSIX equivalent is: /usr/share/tcl8.4/tclConfig.sh
  CYGWIN environment variable option "nodosfilewarning" turns off this
warning.
  Consult the user's guide for more details about POSIX paths:
    http://cygwin.com/cygwin-ug-net/using.html#using-pathnames
Traceback (most recent call last):
  File "setup.py", line 162, in <module>
    if check_for_tk() or (options['build_tkagg'] is True):
  File "/home/Adam/installs/matplotlib-1.0.0/setupext.py", line 816, in
check_for_tk
    explanation = add_tk_flags(module)
  File "/home/Adam/installs/matplotlib-1.0.0/setupext.py", line 1080, in
add_tk_flags
    result = parse_tcl_config(tcl_lib_dir, tk_lib_dir)
  File "/home/Adam/installs/matplotlib-1.0.0/setupext.py", line 938, in
parse_tcl_config
    tk_lib = tk_vars.get("default", "TK_LIB_SPEC")[1:-1].split()[0][2:]
IndexError: list index out of range

I'm trying to display several rectangles (from matplotlib.patches) in 3D. I thought it would be possible to use the add_collection3d method similar to the demo on the examples page:http://matplotlib.sourceforge.net/mpl_toolkits/mplot3d/tutorial.html
However, when I call this method on a PatchCollection, I get an error at line 295 in art3d.py.It expects the collection to have offsets, but offsets is None (by default) when I create the collection. 
/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/mpl_toolkits/mplot3d/art3d.pyc in set_3d_properties(self, zs, zdir)    293     294     def set_3d_properties(self, zs, zdir):--> 295         xs, ys = zip(*self.get_offsets())    296         self._offsets3d = juggle_axes(xs, ys, zs, zdir)    297         self._facecolor3d = self.get_facecolor()
ValueError: need more than 0 values to unpack
I am relatively new to Matplotlib, so I don't know if this is a bug or it is just my own misunderstanding.

Here's a simple example where I try it on a single rectangle and it fails with the above error message.
# ----start example
from mpl_toolkits.mplot3d import Axes3Dfrom matplotlib.collections import PolyCollection,PatchCollectionfrom matplotlib.colors import colorConverterimport matplotlib.pyplot as pltfrom matplotlib.patches import Rectangleimport numpy as np
fig = plt.figure()ax = fig.gca(projection='3d')

rec = Rectangle((5,2),2,2)recs = [rec]col = PatchCollection(recs)ax.add_collection3d(col, zs=[0.6], zdir='z')
# ----end example

Graham 		 	   		  

2010/8/13 Andrea Tomadin <gur...@me...>:
> Dear John,
>
> thank you for your suggestion.  Unfortunately, I have just tried that and I am positive that it does not work.
>
> It works only if the option "frameon" of the figure is True, which is not what I want in this case (I have to produce a pdf with transparent background).

Indeed the code of draw() and resize() do something different.

While draw() only blits to the tk image, resize() creates a new Tk
image.  I believe the bug is real and somewhere buried in tkagg.

For now, you have three options:  Wait until it's fixed in svn, or
monkey-patch using the code of resize().  You can also pass in a dummy
event to .resize() with two attributes, .width and .height, being the
['width'] and ['height'] of the drawing Tk Canvas object, i.e. of your
.canvas._tkcanvas.

Furthermore, does anyone know what the self.resize_event() call in
.resize() is doing?  I'm on 8626.  Looks like a missed
rename-and-back-up error to me.  I don't want to checkout until I'm
familiar with svn to not loose my changes.

Friedrich.

On Mon, Aug 16, 2010 at 2:23 PM, Carlos Anderson <car...@ms...> wrote:

>  Ben,
>
>
>
> Thank you! I removed my current version of matplotlib, then downloaded the
> source code for the newest version, compiled it, installed it, and now
> everything works! I only had one error during compilation, which I fixed by
> downloading and installing libfreetype6-dev.
>
>
Glad to be of help, Carlos.

Ben Root

On 08/16/2010 02:50 PM, Benjamin Root wrote:
> Also, if you never compiled programs before on your system, you might 
> need to install using
> synaptics some build packages.  I think the link lists the 
> dependencies you need to have installed.
As a shortcut, you can install all of the build dependencies for matplotlib using:

  >  apt-get build-deps python-matplotlib

Mike


--
Michael Droettboom
Science Software Branch
Space Telescope Science Institute
Baltimore, Maryland, USA

On Mon, Aug 16, 2010 at 1:17 PM, Carlos Anderson <car...@ms...> wrote:

>  I have an older version of matplotlib (0.98.5.2), but when I do this:
>
>
>
>   sudo apt-get install python-matplotlib
>
>
>
> it says "python-matplotlib is already the newest version" and it doesn't
> download the latest version, which is 1.0.0. How can I get the latest
> matplotlib version using apt-get?
>
>
Carlos,

apt-get uses your linux distribution's repository, which contains the
package as they wish to make available.  Typically, for any particular
version of the OS, they will stick with a particular version of the package.

If you need the latest version of any particular package that is not
available by your distribution yet, you can install it yourself (taking care
to remove the original package) from the source code.  This can be a little
tricky for newcomers, and you will have to be careful when you upgrade to
the next version of your OS and try installing the new version from the
package managers.

To install from source, you might want to start here:

http://matplotlib.sourceforge.net/users/installing.html#installing-from-source

Also, if you never compiled programs before on your system, you might need
to install using synaptics some build packages.  I think the link lists the
dependencies you need to have installed.

I know it can be a little scary if you are a newcomer to linux, but don't be
afraid to ask if you are confused by this.

I hope this helps,
Ben Root

I have an older version of matplotlib (0.98.5.2), but when I do this:

 

  sudo apt-get install python-matplotlib

 

it says "python-matplotlib is already the newest version" and it doesn't
download the latest version, which is 1.0.0. How can I get the latest
matplotlib version using apt-get?

On Sun, Aug 15, 2010 at 10:49 PM, Jae-Joon Lee <lee...@gm...> wrote:
> On Fri, Aug 13, 2010 at 3:34 AM, Jeremy Conlin <jlc...@gm...> wrote:
>> I have a problem with the scaling of the numbers on a colorbar.  The
>> problem occurs when the numbers used as colorbar labels need to be
>> scaled (i.e. by 1E3).  The colorbar correctly puts the scaling value
>> on the top of the colorbar, but instead of of multiplying by a scale
>> factor, addition is used instead.  See the attached script and figure
>> for example.  In the example, it looks like the color scale goes from
>> 0 to 2*5E3.  At least, that's what I thought when I first looked at
>> it.  Instead it means 5000 to 5002.
>>
>> Is there anyway I can scale the colorbar labels by *multiplying* them
>> by a scaling factor instead of *adding* a scaling factor?  That seems
>> more intuitive to me and those I work with.
>>
>
> Given the value range of 5000-5002, I doubt how using the scaling
> factor improve your plot.
>
> To disable the use of offset (+5000),
>
> cb = fig.colorbar(pc)
>
> # do not use offset
> cb.formatter.set_useOffset(False)
>
> cb.update_ticks()
>
> If you do want to use scaling factor,
>
> # to use scaling factor
> cb.formatter.set_scientific(True)
> cb.formatter.set_powerlimits((0,0))
>
> cb.update_ticks()
>
> This will only work with matplotlib v1.0.
> In older versions, try to replace "update_ticks" with
> "update_bruteforce" (but I'm not sure if this will work)
>
> IHTH,
>

Thanks for your suggestions.  I recognize that a range of 5000–5002 is
not much; it was used simply to illustrate my point.

I was able to turn the scaling off with set_useOffset(False).  Is
there anyway to scale by multiplying instead of adding?

Thanks,
Jeremy

On 08/16/2010 06:44 AM, Toby Burnett wrote:
> It is good to know that both issues have been recognized and addressed.
> I searched the bug list and I guess failed to use the right keyword, or
> just didn’t recognize the title.
>
> However, I am not in need of help, and certainly would not do a complete
> reinstall, perhaps introducing new issues, when I now understand how to
> work around both problems. Rather I thought that one of the purposes of
> a user mailing list like this was to alert others, and to provide a
> solution for them. (I spent a good 3 hours tracking it down.) In the
> otherwise excellent documentation, I do not see an easy way to
> understand what bugs have been found in a particular distribution,
> especially the most recent.

We do appreciate bug identification, and especially when it comes with 
an explanation of the bug source and a suggested fix.  You are correct 
that this is a function of the mailing list.

Bugs are reported in two ways: (1) on the mailing list, (2) on the 
sourceforge bug tracker.  In addition, bugs are often identified by a 
developer, and then fixed without any report to the mailing list or the 
tracker. For the latter category, the only ways you can find out whether 
the bug has already been fixed are to search the latest CHANGELOG file 
and the list of commit messages, or--the definitive way--to install and 
test the latest svn version.  Unfortunately, this is not feasible or 
desirable for everyone.

For access to the CHANGELOG and to the commit messages (and to all the 
code and history, if you like), you can use the sourceforge site

http://matplotlib.svn.sourceforge.net/viewvc/matplotlib/trunk/matplotlib/

but I prefer my unofficial mercurial mirror:

http://currents.soest.hawaii.edu/hgstage/hgwebdir.cgi/mpl_hg/

or Andrew's github mirror:

http://github.com/astraw/matplotlib

Eric
>
> --Toby Burnett
>
> *From:* ben...@gm... [mailto:ben...@gm...] *On Behalf Of
> *Benjamin Root
> *Sent:* Monday, August 16, 2010 07:46
> *To:* Toby Burnett
> *Cc:* mat...@li...
> *Subject:* Re: [Matplotlib-users] 1.1.0 changes: bug in
> YAxis.set_ticks_position; new show() behavior
>
> On Sat, Aug 14, 2010 at 11:25 AM, Toby Burnett <tbu...@uw...
> <mailto:tbu...@uw...>> wrote:
>
> YAxis:
> The documentation (http://matplotlib.sourceforge.net/api/axis_api.html)
> says:
>
> "Set the ticks position (left, right, both, default or none) 'both' sets
> the ticks to appear on both positions, but does not change the tick
> labels. 'default' resets the tick positions to the default: ticks on
> both positions, labels at left. 'none' can be used if you don't want any
> ticks. 'none' and 'both' affect only the ticks, not the labels."
>
> To test this:
>
> f=figure()
> plot()
> gca().xaxis.set_ticks_position('none')
> gca().yaxis.set_ticks_position('none')
> f.show()
>
> The y labels, on my 1.0.0 distribution, disappear, because of this line
> of code:
>
> elif position == 'none':
> self.set_tick_params(which='both', right=False, labelright=False,
> left=False, labelleft=False)
> the work-around is to execute that line without turning off the labels.
>
>
> This bug was already found and patched in both the maintenance branch
> and trunk (r8554 and r8555).
>
>     ------------------------------------
>     show()
>
>     This may be an ipython problem, when running it with the -pylab
>     option, the show method now has a Tk.mainloop(), which blocks of
>     course. The work-around it to call the Figure.show() method instead,
>     as above.
>
>     --Toby Burnett
>     University of Washington
>
>
> There have been some recent fixes with Tk (I think) and its behavior
> with regards to .show(). I believe the changes made it to the
> maintenance branch as well. You might want to download the latest
> tarball and install that to fix those problems.
>
> Ben Root
>
>
>
> ------------------------------------------------------------------------------
> 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

It is good to know that both issues have been recognized and addressed. I searched the bug list and I guess failed to use the right keyword, or just didn't recognize the title.

However, I am not in need of help, and certainly would not do a complete reinstall, perhaps introducing new issues, when I now understand how to work around both problems.  Rather I thought that one of the purposes of a user mailing list like this was to alert others, and to provide a solution for them. (I spent a good 3 hours tracking it down.) In the otherwise excellent documentation, I do not see an easy way to understand what bugs have been found in a particular distribution, especially the most recent.

--Toby Burnett

From: ben...@gm... [mailto:ben...@gm...] On Behalf Of Benjamin Root
Sent: Monday, August 16, 2010 07:46
To: Toby Burnett
Cc: mat...@li...
Subject: Re: [Matplotlib-users] 1.1.0 changes: bug in YAxis.set_ticks_position; new show() behavior

On Sat, Aug 14, 2010 at 11:25 AM, Toby Burnett <tbu...@uw...<mailto:tbu...@uw...>> wrote:
YAxis:
The documentation (http://matplotlib.sourceforge.net/api/axis_api.html) says:

"Set the ticks position (left, right, both, default or none) 'both' sets the ticks to appear on both positions, but does not change the tick labels. 'default' resets the tick positions to the default: ticks on both positions, labels at left. 'none' can be used if you don't want any ticks. 'none' and 'both' affect only the ticks, not the labels."

To test this:

f=figure()
plot()
gca().xaxis.set_ticks_position('none')
gca().yaxis.set_ticks_position('none')
f.show()

The y labels, on my 1.0.0 distribution, disappear, because of this line of code:

       elif position == 'none':
           self.set_tick_params(which='both', right=False, labelright=False,
                                left=False, labelleft=False)
the work-around is to execute that line without turning off the labels.

This bug was already found and patched in both the maintenance branch and trunk (r8554 and r8555).

------------------------------------
show()

This may be an ipython problem, when running it with the -pylab option, the show method now has a Tk.mainloop(), which blocks of course. The work-around it to call the Figure.show() method instead, as above.

--Toby Burnett
University of Washington

There have been some recent fixes with Tk (I think) and its behavior with regards to .show().  I believe the changes made it to the maintenance branch as well.  You might want to download the latest tarball and install that to fix those problems.

Ben Root

Those artist in offsetbox module (e.g., TextArea, VPacker,
AnchoredOffsetbox) are mostly a container object, i.e., actual drawing
is done by other artists. And changing the artist attributes of the
container object often does not work.
In your case (I guess you want to make the frame translucent?), the
actual artist responsible for drawing the frame is anchored_box.patch
(==art.patch). So

anchored_box.patch.set_alpha(0.5)

should work.

-JJ


On Tue, Aug 17, 2010 at 12:59 AM, Mathew Yeates <mat...@gm...> wrote:
> Hi
> I want my plot to be visible when I display some text over it. Why
> doesn't this work?
> fig = plt.figure(figsize=(5,5), dpi=100)
> ax=fig.add_axes([0.1,0.1,0.8,.8])
> ax.hist(data)
> box1 = TextArea("Minimum: %f" % mind, textprops=dict(color="k"))
> box2 = TextArea("Maximum: %f" % maxd, textprops=dict(color="k"))
> box = VPacker(children=[box1,box2],
>              align="center",
>              pad=0, sep=5)
> anchored_box = AnchoredOffsetbox(loc=3,
>                                 child=box, pad=0.,
>                                 frameon=True,
>                                 bbox_to_anchor=(0.8, 0.72),
>                                 bbox_transform=ax.transAxes,
>                                 borderpad=0.,
>                                 )
>
>
> art=ax.add_artist(anchored_box)
> art.set_alpha(0.5)
>
> -Mathew
>
> ------------------------------------------------------------------------------
> 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
>

And is NumPy 1.5 compat. confirmed yet?  Thanks!

DG

-- 
Privacy is overrated; Identity isn't.

Thanks, Ben, but I figured out that I just needed to use
frameon=False. This way, I can see teh plot under the text.

Thx

On Mon, Aug 16, 2010 at 9:16 AM, Benjamin Root <ben...@ou...> wrote:
> On Mon, Aug 16, 2010 at 10:59 AM, Mathew Yeates <mat...@gm...>
> wrote:
>>
>> Hi
>> I want my plot to be visible when I display some text over it. Why
>> doesn't this work?
>> fig = plt.figure(figsize=(5,5), dpi=100)
>> ax=fig.add_axes([0.1,0.1,0.8,.8])
>> ax.hist(data)
>> box1 = TextArea("Minimum: %f" % mind, textprops=dict(color="k"))
>> box2 = TextArea("Maximum: %f" % maxd, textprops=dict(color="k"))
>> box = VPacker(children=[box1,box2],
>>              align="center",
>>              pad=0, sep=5)
>> anchored_box = AnchoredOffsetbox(loc=3,
>>                                 child=box, pad=0.,
>>                                 frameon=True,
>>                                 bbox_to_anchor=(0.8, 0.72),
>>                                 bbox_transform=ax.transAxes,
>>                                 borderpad=0.,
>>                                 )
>>
>>
>> art=ax.add_artist(anchored_box)
>> art.set_alpha(0.5)
>>
>> -Mathew
>>
>
> Mathew,
>
> Could you please include a completely self-contained version of the script
> for us to try out?  Plus, which version of matplotlib are you using?
>
> Ben Root
>
>

On Mon, Aug 16, 2010 at 10:59 AM, Mathew Yeates <mat...@gm...>wrote:

> Hi
> I want my plot to be visible when I display some text over it. Why
> doesn't this work?
> fig = plt.figure(figsize=(5,5), dpi=100)
> ax=fig.add_axes([0.1,0.1,0.8,.8])
> ax.hist(data)
> box1 = TextArea("Minimum: %f" % mind, textprops=dict(color="k"))
> box2 = TextArea("Maximum: %f" % maxd, textprops=dict(color="k"))
> box = VPacker(children=[box1,box2],
>              align="center",
>              pad=0, sep=5)
> anchored_box = AnchoredOffsetbox(loc=3,
>                                 child=box, pad=0.,
>                                 frameon=True,
>                                 bbox_to_anchor=(0.8, 0.72),
>                                 bbox_transform=ax.transAxes,
>                                 borderpad=0.,
>                                 )
>
>
> art=ax.add_artist(anchored_box)
> art.set_alpha(0.5)
>
> -Mathew
>
>
Mathew,

Could you please include a completely self-contained version of the script
for us to try out?  Plus, which version of matplotlib are you using?

Ben Root

Hi
I want my plot to be visible when I display some text over it. Why
doesn't this work?
fig = plt.figure(figsize=(5,5), dpi=100)
ax=fig.add_axes([0.1,0.1,0.8,.8])
ax.hist(data)
box1 = TextArea("Minimum: %f" % mind, textprops=dict(color="k"))
box2 = TextArea("Maximum: %f" % maxd, textprops=dict(color="k"))
box = VPacker(children=[box1,box2],
              align="center",
              pad=0, sep=5)
anchored_box = AnchoredOffsetbox(loc=3,
                                 child=box, pad=0.,
                                 frameon=True,
                                 bbox_to_anchor=(0.8, 0.72),
                                 bbox_transform=ax.transAxes,
                                 borderpad=0.,
                                 )


art=ax.add_artist(anchored_box)
art.set_alpha(0.5)

-Mathew

On 8/15/2010 2:06 PM, Paweł Roman wrote:
> Hi,
>
> I have matplotlib 1.0 for python 2.5 (installed from
> matplotlib-1.0.0.win32-py2.5.exe
> <http://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.0/matplotlib-1.0.0.win32-py2.5.exe/download>).
> I'm using windows7 64 bit. When I try to import anything, e.g.
>
> from matplotlib.figure import Figure
>
> I'm getting this error:
>
> Traceback (most recent call last):
>    File "<stdin>", line 1, in <module>
>    File "C:\Python25\Lib\site-packages\matplotlib\figure.py", line 16,
> in <module
>>
>      import artist
>    File "C:\Python25\Lib\site-packages\matplotlib\artist.py", line 6, in
> <module>
>
>      from transforms import Bbox, IdentityTransform, TransformedBbox,
> Transformed
> Path
>    File "C:\Python25\Lib\site-packages\matplotlib\transforms.py", line
> 34, in <mo
> dule>
>      from matplotlib._path import affine_transform
> ImportError: DLL load failed: The specified module could not be found.
>
> When I run this code in python interpreter, I get another clue: windows
> (?) displays the following message in the popup window: The program
> can't start because MSVCP71.dll is missing from your computer. Try
> reinstalling the program to fix this problem.
>
> I have another computer with windows XP 32 bit and it works fine from
> there. I know about other computer with 32 bit system where it also
> works without problems. I've found a ticket in a bug tracker from 2008
> with identical error but no replies. I've downloaded this MSVCP71.dll
> from the internet and put it in windows\system32 but it didnt help
> (still the same error). Seems like the problem with 64 bit windows. But
> is it only me? Can anyone help me with this?
>
> Thanks
>
> Paweł
>

Try to place MSVCP71.dll in C:\Windows\SysWOW64. Windows\system32 is for 
64 bit files only on a 64 bit Windows OS.

--
Christoph

On 8/16/10 5:36 AM, Michael Droettboom wrote:
> For clarification: this change is related to "spines" not "splines".
> Spines are the feature that allows the axes ticks to be plotted
> somewhere in the middle of the plot, rather than always at the edges.
> See here:
>
> http://matplotlib.sourceforge.net/examples/pylab_examples/spine_placement_demo.html?highlight=spine%20codex
>
> Splines would refer to Bezier curves.  See here:
>
> http://matplotlib.sourceforge.net/examples/pylab_examples/dolphin.html?highlight=dolphin%20codex
>
> Which feature are you interested in?
>
> That said, I'm not sure what the spine smart bounds work is about, but I
> suspect Andrew Straw does.


Sorry, I mistyped that: I meant "spines".

We use spines to do the axes in Sage, and in updating the matplotlib 
package, I was wondering if we could take advantage of the "smart bound" 
feature.  It certainly sounded interesting.

Thanks,

Jason
--
Jason Grout

On Sat, Aug 14, 2010 at 11:25 AM, Toby Burnett <tbu...@uw...> wrote:

> YAxis:
> The documentation (http://matplotlib.sourceforge.net/api/axis_api.html)
> says:
>
> "Set the ticks position (left, right, both, default or none) 'both' sets
> the ticks to appear on both positions, but does not change the tick labels.
> 'default' resets the tick positions to the default: ticks on both positions,
> labels at left. 'none' can be used if you don't want any ticks. 'none' and
> 'both' affect only the ticks, not the labels."
>
> To test this:
>
> f=figure()
> plot()
> gca().xaxis.set_ticks_position('none')
> gca().yaxis.set_ticks_position('none')
> f.show()
>
> The y labels, on my 1.0.0 distribution, disappear, because of this line of
> code:
>
>        elif position == 'none':
>            self.set_tick_params(which='both', right=False,
> labelright=False,
>                                 left=False, labelleft=False)
> the work-around is to execute that line without turning off the labels.
>
>
This bug was already found and patched in both the maintenance branch and
trunk (r8554 and r8555).


> ------------------------------------
> show()
>
> This may be an ipython problem, when running it with the -pylab option, the
> show method now has a Tk.mainloop(), which blocks of course. The work-around
> it to call the Figure.show() method instead, as above.
>
> --Toby Burnett
> University of Washington
>
>
There have been some recent fixes with Tk (I think) and its behavior with
regards to .show().  I believe the changes made it to the maintenance branch
as well.  You might want to download the latest tarball and install that to
fix those problems.

Ben Root

For clarification: this change is related to "spines" not "splines".  
Spines are the feature that allows the axes ticks to be plotted 
somewhere in the middle of the plot, rather than always at the edges.  
See here:

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

Splines would refer to Bezier curves.  See here:

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

Which feature are you interested in?

That said, I'm not sure what the spine smart bounds work is about, but I 
suspect Andrew Straw does.

Mike

On 08/14/2010 11:22 AM, Jason Grout wrote:
> Hi all,
>
> First: I noticed in the changelog to 1.0 that a lot of work has gone
> into splines.  Thanks!
>
> In particular, I notice commit message "splines and ticks: implement
> smart bounds" from 20 Dec 2009 (commit
> 58d31c10eef8648cd0fb81d5c0f0bcd0c05fe9a1 in astraw's github repository,
> labeled as git-svn-id:
> https://matplotlib.svn.sourceforge.net/svnroot/matplotlib/trunk/matplotlib@8048
> f61c4167-ca0d-0410-bb4a-bb21726e55ed).
>
> What are "smart bounds"?  It looks desirable to do, but it defaults to
> off, I think.
>
> Thanks,
>
> Jason
>
> ------------------------------------------------------------------------------
> 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
>    


-- 
Michael Droettboom
Science Software Branch
Space Telescope Science Institute
Baltimore, Maryland, USA

On Thu, Aug 12, 2010 at 2:08 PM, Russell E. Owen <ro...@uw...> wrote:
> So...can I convince the automatic sizer to always show the full X (time)
> axis annotations and put all the variable sizing into the data area? Or
> do I have to manually set them somehow?
>

Another option you may try is to use axes_grid1 toolkit.
Attached is a simplified version of the script I have been using.
matplotlib v1.0 is required.

IHTH,

-JJ

On Fri, Aug 13, 2010 at 3:34 AM, Jeremy Conlin <jlc...@gm...> wrote:
> I have a problem with the scaling of the numbers on a colorbar.  The
> problem occurs when the numbers used as colorbar labels need to be
> scaled (i.e. by 1E3).  The colorbar correctly puts the scaling value
> on the top of the colorbar, but instead of of multiplying by a scale
> factor, addition is used instead.  See the attached script and figure
> for example.  In the example, it looks like the color scale goes from
> 0 to 2*5E3.  At least, that's what I thought when I first looked at
> it.  Instead it means 5000 to 5002.
>
> Is there anyway I can scale the colorbar labels by *multiplying* them
> by a scaling factor instead of *adding* a scaling factor?  That seems
> more intuitive to me and those I work with.
>

Given the value range of 5000-5002, I doubt how using the scaling
factor improve your plot.

To disable the use of offset (+5000),

cb = fig.colorbar(pc)

# do not use offset
cb.formatter.set_useOffset(False)

cb.update_ticks()

If you do want to use scaling factor,

# to use scaling factor
cb.formatter.set_scientific(True)
cb.formatter.set_powerlimits((0,0))

cb.update_ticks()

This will only work with matplotlib v1.0.
In older versions, try to replace "update_ticks" with
"update_bruteforce" (but I'm not sure if this will work)

IHTH,

-JJ

> Jeremy
>
> ------------------------------------------------------------------------------
> 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
>
>

Hi all,

I am using matplotlib to create a static graph from a few fairly large
data sets. (Over 500,000 samples)

I know you can pull out datasets of an axes by using the functions
get_xdata() and get_ydata(),
which leads me to wonder how matplotlib handles large data sets? Does
it keep the entire data sets stored in memory?

The application I am trying develop I hope to be able to run on low
end PC's with little RAM.

If data sets are indeed stored in RAM, is there a way to discard the
plot data after a plot had been created to leave just a static image?
Or, read and store data points directly from a file?

Thanks for your help.

Craig