matplotlib-users Mailing List for matplotlib (Page 7)

On 2008年3月01日, Francesco Biscani apparently wrote:
> I'm currently working mostly in C++ and exploring 
> integration with Python through Boost.Python+IPython+MPL. 
> I enjoy working in Python, but I'm afraid of making a more 
> consistent switch mainly for fear of losing the static 
> type checking that C++ gives me. I have the horrifying 
> feeling that if I were to write much code in Python 
> I could break it in so many ways just because of this, and 
> I'd have the constant perception of not having my back 
> covered by the compiler, at least for this kind of errors. 
I suspect there is no answer to this fear except trying it.
If you want to hear your fear ridiculed, you can post this
query on comp.lang.python. If you want to hear your fear
lauded for its sensibility, post it to comp.lang.c++.
That said, note that Python supports properties (getters and 
setters) that can do runtime checking.
Also Python 3 supports function annotations, which someone 
will certainly exploit to provide type checking. 
I guess you might find PEP 3119 (ABCs) relevant as well.
Cheers,
Alan Isaac

On Fri, Mar 21, 2008 at 2:00 AM, Eric Firing <ef...@ha...> wrote:
> Charlie Moad does the Windows releases. I don't know what compiler he uses.
>
> It would be nice if some people who have successfully built on Windows
> could collectively assemble a step-by-step account of how to go from a
> bare Win box to a working mpl (preferably compiled with mingw); but
> maybe this would take more effort than it is worth. I am on shaky
> ground suggesting it, because it is not something I can help with at
> all, and I don't even have a clear picture of what it would require. I
> gather a similar account would be useful for OS X.
I made fairly detailed notes when I build numpy, scipy and mpl on a
pristine powerbook:
http://ipython.scipy.org/moin/Py4Science/InstallationOSX
Unfortunately, I am having the same compiler error that Zachary and
others are discussing in another thread (on the trunk but not the
branch)
JDH

On Tue, Feb 26, 2008 at 3:23 AM, Michaël Douchin
<mic...@la...> wrote:
> What is the good way to do what I reach for ?
> I have much trouble to know how to define the c parameter of scatter. I
> saw the "arrange" thing somewhere in internet, but don't understand what
> it is for.
Have you tried the scatter demos? See for example
http://matplotlib.sf.net/examples/scatter_demo2.py
JDH

On Fri, Mar 21, 2008 at 2:34 PM, Kenneth Miller <xke...@gm...> wrote:
> So i've experimented with pcolor, and it's not really what I'm looking
> for. What I'm attempting to do is plot some XY coordinates, just a
> simple function, with the line being colored differently depending on
> the value of that function. So if perhaps you were plotting
> temperature vs time, you'd see the line change to red when it
> approached higher temperatures and blue when it approached lower
> temperatures.
The "multi-colored line" recipe should be enough to get you started:
 http://www.scipy.org/Cookbook/Matplotlib/MulticoloredLine
JDH

On Sat, Feb 23, 2008 at 3:25 AM, pmarsh <pm...@ha...> wrote:
> Hello List,
>
> I am using python-matplotlib version 0.87.7-0.3ubuntu1(feisty) and
> having some problems with a script that uses it. I get this import error
> and am not sure how to debug it further, any assistance would be greatly
> appreciated.
It looks like the basemap version you are using is assuming a more
recent version of matplotlib. You will probably either need to
upgrade to more recent matplotlib (0.91.2 is the most recent) or
downgrade to an older basemap.
JDH

On Thu, Mar 20, 2008 at 7:32 PM, carlwenrich <car...@ya...> wrote:
>
> I've searched the user manual (and this forum) but I don't see anything that
> helps.
Assuming you mean what we call the tick labels:
import matplotlib
matplotlib.rcParams['xtick.labelsize'] = 14
matplotlib.rcParams['ytick.labelsize'] = 14
These can also be changed in your matplotlibrc file -- see
http://matplotlib.sf.net/matplotlibrc
JDH

All,
A quick question....
So i've experimented with pcolor, and it's not really what I'm looking 
for. What I'm attempting to do is plot some XY coordinates, just a 
simple function, with the line being colored differently depending on 
the value of that function. So if perhaps you were plotting 
temperature vs time, you'd see the line change to red when it 
approached higher temperatures and blue when it approached lower 
temperatures.
Any advice as to how to do this?
Regards,
Ken

On Mar 21, 2008, at 6:12 AM, Michael Droettboom wrote:
> I vaguely recall a bug whereby mathtext on PDF was upside down 
> (because the direction of the y-axis was not being inverted)... but 
> I can't find the bug report.
>
> It does seem to work in 0.90.1 and 0.91.2 (on Linux at least). Are 
> you able to upgrade?
Hm.
On my Linux box, Well, easy_install should upgrade me to 91.2, but 
won't build because ft2build.h is missing...? Apparently that's part 
of freetype v2, which I have installed...
On my Mac, easy_install says that matplotlib 0.87.7 is the active 
version and the best version.
I guess I can't easy install...
02:46 PM t:/home/simsong# easy_install matplotlib
Searching for matplotlib
Reading http://pypi.python.org/simple/matplotlib/
Reading http://matplotlib.sourceforge.net
Reading http://sourceforge.net/project/showfiles.php?group_id=80706&package_id=82474
Reading http://sourceforge.net/project/showfiles.php?group_id=80706
Best match: matplotlib 0.91.2
Downloading http://downloads.sourceforge.net/matplotlib/matplotlib-0.91.2.tar.gz?modtime=1199627250&big_mirror=0
Processing matplotlib-0.91.2.tar.gz
Running matplotlib-0.91.2/setup.py -q bdist_egg --dist-dir /tmp/ 
easy_install-YMu8YK/matplotlib-0.91.2/egg-dist-tmp-AOR3hR
= 
= 
= 
= 
========================================================================
BUILDING MATPLOTLIB
 matplotlib: 0.91.2
 python: 2.4.4 (#1, Oct 23 2006, 13:58:18) [GCC 4.1.1
 20061011 (Red Hat 4.1.1-30)]
 platform: linux2
REQUIRED DEPENDENCIES
 numpy: 1.0.3
 freetype2: found, but unknown version (no pkg-config)
 * WARNING: Could not find 'freetype2' headers 
in any
 * of '/usr/local/include', '/usr/include', '.',
 * '/usr/local/include/freetype2',
 * '/usr/include/freetype2', './freetype2'.
OPTIONAL BACKEND DEPENDENCIES
 libpng: found, but unknown version (no pkg-config)
 * Could not find 'libpng' headers in any of
 * '/usr/local/include', '/usr/include', '.'
 Tkinter: no
 * Tkinter present, but header files are not 
found.
 * You may need to install development packages.
 wxPython: no
 * WXAgg's accelerator requires `wx-config'. 
The
 * `wx-config' executable could not be located 
in any
 * directory of the PATH environment variable. 
If you
 * want to build WXAgg, and wx-config is in some
 * other location or has some other name, set 
the
 * WX_CONFIG environment variable to the full 
path of
 * the executable like so: export WX_CONFIG=/ 
usr/lib
 * /wxPython-2.6.1.0-gtk2-unicode/bin/wx-config
 Gtk+: no
 * pygtk present but import failed
 Qt: Qt: 3.3.7, PyQt: 3.17
 Qt4: no
 Cairo: 1.2.6
OPTIONAL DATE/TIMEZONE DEPENDENCIES
 datetime: present, version unknown
 dateutil: present, version unknown
 pytz: 2006p
OPTIONAL USETEX DEPENDENCIES
 dvipng: 1.5
 ghostscript: 8.15.4
 latex: 3.141592
 pdftops: 3.00
EXPERIMENTAL CONFIG PACKAGE DEPENDENCIES
 configobj: matplotlib will provide
 enthought.traits: matplotlib will provide
[Edit setup.cfg to suppress the above messages]
= 
= 
= 
= 
========================================================================
warning: no files found matching 'NUMARRAY_ISSUES'
warning: no files found matching 'MANIFEST'
warning: no files found matching 'matplotlibrc'
warning: no files found matching 'lib/matplotlib/toolkits'
no previously-included directories found matching 'examples/_tmp_*'
In file included from src/ft2font.cpp:2:
src/ft2font.h:11:22: error: ft2build.h: No such file or directory
src/ft2font.h:12:10: error: #include expects "FILENAME" or <FILENAME>
src/ft2font.h:13:10: error: #include expects "FILENAME" or <FILENAME>
src/ft2font.h:14:10: error: #include expects "FILENAME" or <FILENAME>
src/ft2font.h:15:10: error: #include expects "FILENAME" or <FILENAME>
src/ft2font.h:16:10: error: #include expects "FILENAME" or <FILENAME>
src/ft2font.h:31: error: ‘FT_Bitmap’ has not been declared
src/ft2font.h:31: error: ‘FT_Int’ has not been declared
src/ft2font.h:31: error: ‘FT_Int’ has not been declared
src/ft2font.h:75: error: expected ‘,’ or ‘...’ before ‘&’ token
src/ft2font.h:75: error: ISO C++ forbids declaration of ‘FT_Face’ with 
no type
src/ft2font.h:81: error: expected ‘,’ or ‘...’ before ‘&’ token
src/ft2font.h:81: error: ISO C++ forbids declaration of ‘FT_Face’ with 
no type
src/ft2font.h:121: error: ‘FT_Face’ does not name a type
src/ft2font.h:122: error: ‘FT_Matrix’ does not name a type
src/ft2font.h:123: error: ‘FT_Vector’ does not name a type
src/ft2font.h:124: error: ‘FT_Error’ does not name a type
src/ft2font.h:125: error: ‘FT_Glyph’ was not declared in this scope
src/ft2font.h:125: error: template argument 1 is invalid
src/ft2font.h:125: error: template argument 2 is invalid
src/ft2font.h:126: error: ‘FT_Vector’ was not declared in this scope
src/ft2font.h:126: error: template argument 1 is invalid
src/ft2font.h:126: error: template argument 2 is invalid
src/ft2font.h:133: error: ‘FT_BBox’ does not name a type
src/ft2font.cpp:43: error: ‘FT_Library’ does not name a type
src/ft2font.cpp:92: error: variable or field ‘draw_bitmap’ declared void
src/ft2font.cpp:92: error: ‘int FT2Image::draw_bitmap’ is not a static 
member of ‘class FT2Image’
src/ft2font.cpp:92: error: ‘FT_Bitmap’ was not declared in this scope
src/ft2font.cpp:92: error: ‘bitmap’ was not declared in this scope
src/ft2font.cpp:93: error: ‘FT_Int’ was not declared in this scope
src/ft2font.cpp:94: error: ‘FT_Int’ was not declared in this scope
src/ft2font.cpp:94: error: initializer expression list treated as 
compound expression
src/ft2font.cpp:94: error: expected ‘,’ or ‘;’ before ‘{’ token
error: Setup script exited with error: command 'gcc' failed with exit 
status 1
Exception exceptions.OSError: (2, 'No such file or directory', 'src/ 
image.cpp') in <bound method CleanUpFile.__del__ of 
<setupext.CleanUpFile instance at 0xb806c8>> ignored
Exception exceptions.OSError: (2, 'No such file or directory', 'src/ 
transforms.cpp') in <bound method CleanUpFile.__del__ of 
<setupext.CleanUpFile instance at 0xb7fb00>> ignored
Exception exceptions.OSError: (2, 'No such file or directory', 'src/ 
backend_agg.cpp') in <bound method CleanUpFile.__del__ of 
<setupext.CleanUpFile instance at 0xb80560>> ignored
02:46 PM t:/home/simsong#
02:49 PM t:/home/simsong# yum list | grep -i freetype
freetype.i386 2.2.1-17.fc6 
installed
freetype.x86_64 2.2.1-17.fc6 
installed
freetype-demos.x86_64 2.2.1-17.fc6 updates
freetype-devel.i386 2.2.1-17.fc6 updates
freetype-devel.x86_64 2.2.1-17.fc6 updates
freetype1.x86_64 1.4-0.2.pre.fc6 extras
freetype1-devel.x86_64 1.4-0.2.pre.fc6 extras
freetype1-utils.x86_64 1.4-0.2.pre.fc6 extras
02:49 PM t:/home/simsong#

I'd guess PNG won't get much smaller because you have a lot of different
colored pixels. PNG compresses most when you have a sparser plot. I'd
suggest that you try using JPG. It will compress the multi-colored portion
of your plot way down. You may see a few artifacts if you look carefully at
the axes: instead of pixels white,black,white you might get
white,grey,black,grey,white.
> -----Original Message-----
> From: mat...@li...
> [mailto:mat...@li...] On Behalf Of
> Einar M. Einarsson
> Sent: Monday, March 17, 2008 5:08 AM
> To: mat...@li...
> Subject: [Matplotlib-users] PNG filesize
> 
> 
> Hi all,
> 
> I'm trying to find ways to make the file-size of my PNG images smaller.
> 
> When I generate my 660*440px image I get a big 168kb file.
> (8bit RGB color model, has an alpha channel (need that) but no
> interlacing scheme)
> 
> Here it is:
> http://metphys.org/eme/T05.png
> 
> I'm using the savefig method of-course.
> 
> 
> To see how much I could compress it I used pngcrush (the best tool
> according to the interwebs) and got it down to 128kb.
> 
> But thats still way to large for my intended use. (plotting results
> from an operational weather model, see. www.belgingur.is
> We are currently using IDL.)
> 
> From what I've read about PNG files, which is supposed to be rather
> compact image format, it seems to me that the most effective way is
> to have an indexed color table.
> 
> So to cut it short:
> 
> Is there any way to save a PNG file with an indexed color table?
> 
> Or do you see any other way to shrink the files?
> 
> 
> Best regards.
> Einar M. Einarsson
> www.belgingur.is
> 
> -----------------------------------------------------------------------
> --
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Last Thursday I did a tutorial on matplotlib for PyCon, and prepared a
couple of new tutorials for the workshop. I've attached these here in
their ReST source as in PDF for those of you interested in learning a
little more about the matplotlib API. The source and accompanying
code that generates the figures are found in the "doc" subdirectory of
the svn trunk, so feel free to make changes, bugfixes and submit
patches or new tutorials.
The first tutorial is an overview of the matplotlib Artist API and the
second is on event handling and object picking.
JDH

Chris Withers wrote:
> Michael Droettboom wrote:
>> The "backend_driver.py" example runs a number of other examples using 
>> a number of different backends. That's been a reasonably successful 
>> set of regression and coverage tests. It's not perfect, but it's a 
>> start. There are also some lower-level unit tests for 
>> frequently-reoccurring bugs in the unit directory.
>
> Indeed, is there any kind of "full unit test" suite that a developer 
> can run when changing things ot make sure they haven't fubarred anything?
>
backend_driver.py is, AFAIK, the closest thing to that. I recently did 
a coverage analysis of it (with coverage.py), and it hits something like 
98% of the non-error case code, so it's pretty good.
What we don't have is an automated regression test framework to see if 
the results remain correct. There are a number of reasons why this is a 
non-trivial task -- there was a thread about that on this list a few 
months ago. Unfortunately, I can't find it... Maybe someone else 
remembers the subject line?
Mike
-- 
Michael Droettboom
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA

Kenneth Miller wrote:
> 
> back in time. When i pass plot_dates timestamps for the y axis, and 
> integers for the x axis it simply displays the y-axis as floats.
did you try:
plot_dates(x,dates,ydate=True)
?
Chris
-- 
Simplistix - Content Management, Zope & Python Consulting
 - http://www.simplistix.co.uk

On Mar 21, 2008, at 11:56 AM, Chris Withers wrote:
> Kenneth Miller wrote:
>> Is it possible to plot dates on the Y-axis? I'd like to have 
>> dates on the y axis descending or ascending versus my values on the 
>> x - axis. Is it possible to do this or simply switch the axis?
>
> Not sure what you mean, have you just tried it with plot or 
> plot_dates?
A common type of graph in my work is to see the value (from 0 to 
something) represented on the x axis, and instead of the y 
representing a value, it represents time. The closer the Y value gets 
to the x-axis, the closer it is to the current time, and the further 
away the further back in time. When i pass plot_dates timestamps for 
the y axis, and integers for the x axis it simply displays the y-axis 
as floats.
Regards,
Kenneth Miller
>
>
> What problems are you experiencing?
>
> cheers,
>
> Chris
>
> -- 
> Simplistix - Content Management, Zope & Python Consulting
> - http://www.simplistix.co.uk

On Fri, Mar 21, 2008 at 11:27 AM, Kenneth Miller <xke...@gm...> wrote:
> All,
>
> Is it possible to plot dates on the Y-axis? I'd like to have
> dates on the y axis descending or ascending versus my values on the x
> - axis. Is it possible to do this or simply switch the axis?
Not a problem -- with recent versions of mpl, you can simply pass a
sequence of date objects directly. Use Axes.invert_yaxis to change
the order from ascending to descending. Here is an example loading
some data from a CSV file and plotting floats on the x axis and dates
on the y axis
In [7]: import matplotlib.mlab as mlab
In [8]: r = mlab.csv2rec('aapl.csv')
In [9]: r.sort()
In [10]: r[-5:]
Out[10]:
recarray([ (datetime.date(2008, 2, 11), 128.00999999999999,
129.97999999999999, 127.2, 129.44999999999999, 42886900,
129.44999999999999),
 (datetime.date(2008, 2, 12), 130.69999999999999, 131.0, 123.62,
124.86, 43749900, 124.86),
 (datetime.date(2008, 2, 13), 126.68000000000001, 129.78,
125.63, 129.40000000000001, 34542300, 129.40000000000001),
 (datetime.date(2008, 2, 14), 129.40000000000001,
130.80000000000001, 127.01000000000001, 127.45999999999999, 34074900,
127.45999999999999),
 (datetime.date(2008, 2, 15), 126.27, 127.08, 124.06, 124.63,
32163400, 124.63)],
 dtype=[('date', '|O4'), ('open', '<f8'), ('high', '<f8'),
('low', '<f8'), ('close', '<f8'), ('volume', '<i4'), ('adj_close',
'<f8')])
In [11]: fig = figure()
In [12]: ax = fig.add_subplot(111)
In [13]: ax.plot(r.close, r.date, 'o')
Out[13]: [<matplotlib.lines.Line2D instance at 0x237fc10>]
In [14]: draw()
In [15]: ax.invert_yaxis()
In [16]: draw()

Andrew Charles wrote:
> Looking back over the easy_install output it looks as if it does
> download another tarball and try to build it. I read another thread
> where this was happening to someone else.
What did you try to install? was it this:
matplotlib-0.91.2-py2.5-macosx-10.3-fat.egg
which I think should work, so if it doesn't, maybe we can figure out why 
not. It worked for me with MacPython2.5 on OS-X 10.4
Charlie, am I right that that should work?
> I'll look at other libJPEGs - if that doesn't make it fly I might need
> to take your other advice and nuke macpython. I'm a little hesitant
> about over-writing the system installed numpy though.
I agree. MacPython is a good way to go, and we should have a binary for 
it, so maybe we can solve this. I don't have Leopard, so I'm not sure I 
can help much though.
-Chris
-- 
Christopher Barker, Ph.D.
Oceanographer
Emergency Response Division
NOAA/NOS/OR&R (206) 526-6959 voice
7600 Sand Point Way NE (206) 526-6329 fax
Seattle, WA 98115 (206) 526-6317 main reception
Chr...@no...

Michael Droettboom wrote:
> The "backend_driver.py" example runs a number of other examples using a 
> number of different backends. That's been a reasonably successful set 
> of regression and coverage tests. It's not perfect, but it's a start. 
> There are also some lower-level unit tests for frequently-reoccurring 
> bugs in the unit directory.
Indeed, is there any kind of "full unit test" suite that a developer can 
run when changing things ot make sure they haven't fubarred anything?
cheers,
Chris
-- 
Simplistix - Content Management, Zope & Python Consulting
 - http://www.simplistix.co.uk

Pierre GM wrote:
> Your data is indexed in time, right ? Your x-axis is a date object ? Then use 
> scikits.timeseries
> http://scipy.org/scipy/scikits/wiki/TimeSeries
I'm not sure what this is giving me.
The dates are all python datetimes in a list already.
The missing values started off as '', I turned those into nan and then 
created a ma with the nan's masked.
What more would TimeSeries give me?
> the link above. I must admit we didn't implement poly_between for timeseries. 
> Most likely, we'd have to implement it for regular masked arrays first, as 
> mentioned by Eric.
OK.
> What you could do is to fill your array with some kind of baseline, such as 0, 
> or your minimum data, or wtvr. That's just a quick trick and no fix.
Indeed, that's what I had to do.
I have to admit, I see some interesting things while scanning that wiki 
page, but nothing that would have helped me...
cheers,
Chris (who might well be missing something...)
-- 
Simplistix - Content Management, Zope & Python Consulting
 - http://www.simplistix.co.uk

Eric Firing wrote:
> It would be nice if some people who have successfully built on Windows 
> could collectively assemble a step-by-step account of how to go from a 
> bare Win box to a working mpl (preferably compiled with mingw); but 
> maybe this would take more effort than it is worth.
I don't think so -- that would be great!
 > I gather a similar account would be useful for OS X.
It would, and it's been done at various times by various people (myself 
included, a good while back). I don't know if there is a recent one out 
there.
both of these should be in an easy-to-find place on the Wiki.
As for OS-X: OS-X is a pain in the &^%^ because Apple doesn't include 
all the libs MPL needs (I'm not up to date on this, but I think libpng, 
libjpeg and libfreetype). On Windows, this is also the case, but I 
understand that the MPL distro includes them. On OS-X, it is just easy 
enough to get them elsewhere that there isn't the motivation to get them 
included in MPL. However, the problem comes in that there are way too 
many ways to get these libs on OS-X:
 - compile from a tarball
 - fink
 - macports
 - various binary installers.
Also, even with these methods, there are issues of how they compiled: 
Intel, PPC, or Universal.
Ideally, we'd all use the same Universal libs, but the fact is that it 
it easier to get it running on your own machine by using fink or macport 
or the tarball, and building just for your system, but then you dont' 
get something fully redistributable.
Further complicating all this is that there are way too many versions of 
Python for OS-X: Framework, fink, macports, activestate, Apple's, 
"MacPython", etc.
Personally, I'd really like there to be a decision about what is 
supported by MPL, and I think that should be:
MacPython2.5 (for OS-X 10.3.9 to 10.5)
Universal Builds
Extra libs statically linked in.
Most recent numpy release.
Tk and wx back-ends
Cocoa? (is that functional?)
It's a bit of pain to get set up to build this, but once you're set up, 
it's easy to do and distribute it an everyone can use it.
Also, I THINK it's possible to build a binary distro (egg?) that will 
work with both MacPython2.5 and Apple's 25 (which comes with OS-X 10.5). 
Can anyone confirm this?
Either easy_install or *.dmgs (build with bdist_mpkg) that can be 
downloaded and clicked on would be fine.
The wxPython project supports OS-X builds like this, and it works great.
fink and macports can do their own thing.
Is there currently an appropriate place in the Wiki for "How to install 
on OS-X" ?
After all that writting, I see:
 	matplotlib-0.91.2-py2.5-macosx-10.3-fat.egg
on the sourceforge download site -- this looks like it's just what I'm 
suggesting. If that didn't work for the OP, we should figure out why not.
-Chris
-- 
Christopher Barker, Ph.D.
Oceanographer
Emergency Response Division
NOAA/NOS/OR&R (206) 526-6959 voice
7600 Sand Point Way NE (206) 526-6329 fax
Seattle, WA 98115 (206) 526-6317 main reception
Chr...@no...

Kenneth Miller wrote:
> Is it possible to plot dates on the Y-axis? I'd like to have 
> dates on the y axis descending or ascending versus my values on the x 
> - axis. Is it possible to do this or simply switch the axis?
Not sure what you mean, have you just tried it with plot or plot_dates?
What problems are you experiencing?
cheers,
Chris
-- 
Simplistix - Content Management, Zope & Python Consulting
 - http://www.simplistix.co.uk

All,
 Is it possible to plot dates on the Y-axis? I'd like to have 
dates on the y axis descending or ascending versus my values on the x 
- axis. Is it possible to do this or simply switch the axis?
Thanks!
Regards,
Ken

Chris:
I have used cygwin to compile numpy and scipy svn versions for win XP on 
my old athlon system. I believe the scipy site has some detailed 
instructions. However, I have done a lot of compiling on my linux 
system and so that part was easy (configure, make, make install, etc) I 
don't know about mpl but scipy for sure was very sensitive to the 
platform and the stock scipy distro blew up big time doing curve fitting 
optimizations. Compiling my own versions made those problems go away, 
thankfully. You can download cygwin and install it easily enough but 
then you need to pick which of the included programs to install and that 
is a pain unless you have disk space and can do everything. 
For numpy and scipy I did the following in the cygwin prompt after 
cd-ing to where I keep my sources:
 "/cygdrive/c/python25/python.exe setup.py config --compiler=mingw32 
build \
 --compiler=mingw32 bdist_wininst " and this produced a windows .exe 
installer.
I tried mpl last night and it failed due to some dependencies, I think 
it was freetype that was missing and I couldn't satisfy the requirements 
but maybe someone on the list can enlighten us. With that taken care 
of, it should be straightforward.
Wayne
Chris Withers wrote:
> Michael Droettboom wrote:
> 
>>> That's cool'n'all, but when is svn going to make it into a Windows 
>>> binary release? ;-)
>>> 
>>> 
>> I suspect your question is somewhat rhetorical, but... it will probably 
>> be a while ;) 
>> 
>
> Why is that? Who cranks out the binary releases on Windows and what 
> compiler do they use?
>
> 
>> I know a lot of people (myself included) have had success with MinGW. 
>> 
>
> What's the "official" compiler used, though?
>
> 
>> It's a good learning experience, and there's lots on this list willing 
>> to help. If we can get more SVN Windows users on board, more crazy 
>> Windows-only bugs will get found and squashed sooner... ;)
>> 
>
> Well, tell me how to get the svn trunk and how to compile and I'll give 
> it a go :-)
>
> cheers,
>
> Chris
>
> 

I vaguely recall a bug whereby mathtext on PDF was upside down (because 
the direction of the y-axis was not being inverted)... but I can't find 
the bug report.
It does seem to work in 0.90.1 and 0.91.2 (on Linux at least). Are you 
able to upgrade?
Cheers,
Mike
Simson Garfinkel wrote:
> Hi. I'm making scientific graphs again, so i just got back on this 
> mailing list after a two year hiatus.
>
> I am using:
> matplotlib-0.90.0-py2.5-macosx-10.4-fat.egg
>
> Here is a simple program:
>
> #!/usr/bin/python
> #
> # Do our plots with matplotlib
>
> import matplotlib
> matplotlib.use('Agg')
> from pylab import *
>
> fig = figure(num=1,figsize=(6.5,4))
> ax = fig.add_axes([.15,.3,.8,.65])
> ax.set_title("Why is the Y axes upside down?")
> ax.set_yscale('log')
> ax.set_ylabel('Queries per second')
>
> ax.bar(1,1000,.5,color='r')
> ax.bar(2,5000,.5,color='b')
> ax.bar(3,55000,.5,color='g')
>
> fig.savefig("broken_log_demo.png")
>
> Here is the plot with a PNG:
>
>
>
> ------------------------------------------------------------------------
>
>
>
>
> Well, that looks good.
>
> But here is the plot when I change the type to PDF:
>
>
>
>
> Notice the the axes are reversed.
>
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> ------------------------------------------------------------------------
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
-- 
Michael Droettboom
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA

Chris Withers wrote:
> Michael Droettboom wrote:
>>> That's cool'n'all, but when is svn going to make it into a Windows 
>>> binary release? ;-)
>>> 
>> I suspect your question is somewhat rhetorical, but... it will 
>> probably be a while ;) 
>
> Why is that? Who cranks out the binary releases on Windows and what 
> compiler do they use?
Matplotlib recently underwent a major refactoring of the code to make it 
easier to add new transformations and projections to the system. This 
code is significantly less tested than the latest release. The idea of 
a release is that it is reasonably well-tested and reasonably stable. 
Obviously, the matplotlib releases aren't perfect, but I suspect 0.91.2 
is far more safe and predictable than SVN trunk is right now.
However, there is a branch, matplotlib_0_91_maint, that *theoretically* 
should be the latest release + any important bugfixes, but no new 
features. I say *theoretically* because occasionally bugfixes are made 
only to the trunk. If you find such a fix, let us know and hopefully 
someone can backport it.
Mike
-- 
Michael Droettboom
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA

Chris Withers wrote:
> Eric Firing wrote:
> 
>> If you are referring to scripts in the matplotlib/examples/ subdirectory 
>> then you must have a version in which some of those scripts had not been 
>> brought up to date with the rest of matplotlib. 
>> 
>
> You should turn them into unit tests as well as examples.
> 
The "backend_driver.py" example runs a number of other examples using a 
number of different backends. That's been a reasonably successful set 
of regression and coverage tests. It's not perfect, but it's a start. 
There are also some lower-level unit tests for frequently-reoccurring 
bugs in the unit directory.
Mike
-- 
Michael Droettboom
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA

On Fri, Mar 21, 2008 at 10:13 AM, Christopher Barker
<Chr...@no...> wrote:
> > Compiling Matplotlib from source, or easy_installing the egg
>
>
> >>From the egg:
> > ld: in /sw/lib/libJPEG.dylib, file is not of required architecture for
>
> hmm -- odd, I wouldn't think the egg should be linked against what looks
> like a macports libJPEG -- are you sure it isn't trying to build the
> egg when you easy_install?
Looking back over the easy_install output it looks as if it does
download another tarball and try to build it. I read another thread
where this was happening to someone else.
I'll look at other libJPEGs - if that doesn't make it fly I might need
to take your other advice and nuke macpython. I'm a little hesitant
about over-writing the system installed numpy though.
Andrew