matplotlib-users Mailing List for matplotlib (Page 5)

On Tuesday 24 January 2006 16:36, Stephen Walton wrote:
> Hi,
>
> I've been testing my matplotlib 0.86.2 installation on Fedora Core 4 and
> hit a problem I can't solve. If I do "python tex_demo.py" in the
> examples directory it seems to hang up. A quick "ps faxw" and a bit of
> digging shows that the problem is that on my installation of tetex, TeX
>
> isn't recognizing \rmfamily:
> > freyer:~/.matplotlib/tex.cache> tex 0885e19fbc15d1bc2b5e1d3ac6018bee.tex
> > This is TeX, Version 3.141592 (Web2C 7.5.4)
> > (./0885e19fbc15d1bc2b5e1d3ac6018bee.tex
> > ! Undefined control sequence.
> > l.5 {\rmfamily
> > \TeX\ is Number
> > $\displaystyle\sum_{n=1}^\infty\frac{-e^{i\pi...
> >
> > ? x
>
> What is the problem here? I searched the mailing list and no one else
> seems to be having this trouble, so obviously it's my fault, but I can't
> quite figure out how.
I wonder what \rmfamily is doing there, I don't see it in my copy of 
tex_demo.py. It works ok for me even if I add \rmfamily if I have 
text.tex.engine=latex in matplotlibrc, and fails if I have it set to tex. I 
I'm guessing this is the source of your trouble.
Darren
P.S. I'm not able to run tex_demo at all if I use the tex backend instead of 
latex. I see that some changes were made to the font commands in 
texmanage.get_tex_command, have these changes been tested with the tex 
backend?

Ryan Krauss wrote:
> I have actually read the AFPL as a way to try and understand the
> difference. I think the only difference between the AFPL and GPL is
> that the GPL allows you to charge someone for the service of
> distributing a copy of the software where the AFPL does not. (I could
> sell CD's of GPL ghostscript if I wanted to, but anyone who bought
> them must be free to give away copies to whoever they want. I don't
> know if the AFPL would allow me to even charge for the media cost.)
> 
> I have no idea why this small difference has lead to at least two
> different branches of ghostscript.
Aladdin wanted to make money by selling Ghostscript but still play sorta nice
with the open source world. So when they make a stable release, it is under the
AFPL for N years, then it reverts to the GPL. I think the FSF may have
officially forked the project, too, but I'd have to reread the webpages to be
sure. And I don't actually care enough to do so. :-)
-- 
Robert Kern
rob...@gm...
"In the fields of hell where the grass grows high
 Are the graves of dreams allowed to die."
 -- Richard Harter

Hi,
> I have no idea why this small difference has lead to at least two
> different branches of ghostscript.
I think this:
http://www.artifex.com/licensing/
was the point. In other words, AFPL ghostscript is dual licenced...
Best,
Matthew

I have actually read the AFPL as a way to try and understand the
difference. I think the only difference between the AFPL and GPL is
that the GPL allows you to charge someone for the service of
distributing a copy of the software where the AFPL does not. (I could
sell CD's of GPL ghostscript if I wanted to, but anyone who bought
them must be free to give away copies to whoever they want. I don't
know if the AFPL would allow me to even charge for the media cost.)
I have no idea why this small difference has lead to at least two
different branches of ghostscript.
Ryan
On 1/24/06, Fernando Perez <Fer...@co...> wrote:
> Stephen Walton wrote:
> > Fernando Perez wrote:
> >
> >
> >>cd /your/gsdownload/dir # I got 8.50, the current gnu one.
> >
> >
> > There are two parallel versions of ghostscript about whose relationship
> > I'm unclear. Version 8.50+ comes from www.ghostscript.com and is calle=
d
> > "AFPL Ghostscript". Fedora ships with the one from the GNU project, is
> > at version 8.16 and is obtained via ftp://ftp.gnu.org/ftp/ghostscript.
>
> It _is_ confusing. I used 'GPL ghostscript' from:
>
> http://www.cs.wisc.edu/~ghost/
>
> version 8.50 as listed on that page. I haven't the foggiest idea how tha=
t
> number maps to gnu-ghostscript, CUPS-ghostscript, or anything else.
>
> All I know is that I got mpl to shut up after installing this :)
>
> Cheers,
>
> f
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through log fi=
les
> for problems? Stop! Download the new AJAX search engine that makes
> searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D103432&bid=3D230486&dat=
=3D121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Dear all,
I am just looking at matplotlib for the first time and liking very much
what I see. I have a requirement to try and plot a horizontal scan from
a radar. The incoming data is simply array(distanceFromInstrument,
azimuth). Has anybody tried doing for a polar plot without converting
the data to a lat/lon grid first?
An example of what I'd like to do is visible at:
http://www.chilbolton.rl.ac.uk/weather/latest/latest_ppi_v2.png
Thanks,
Ag
--=20
Ag Stephens				 Tel: +44 (0)1392
884263=20
BADC - UK Met Office Coordinator	 Fax: +44 (0)1392 885681
Met Office, A2 - W004, FitzRoy Road, Exeter, EX1 3PB, UK.
E-mail (Met Office):	ag....@me...
E-mail (BADC):		a.s...@rl...
Home page:		 http://home.badc.rl.ac.uk/astephens
 =20
BADC details: =20
British Atmospheric Data Centre, Rutherford Appleton Laboratory,
Chilton, Didcot, OX11 0QX, UK. 	http://badc.nerc.ac.uk

Stephen Walton wrote:
> Fernando Perez wrote:
> 
> 
>>cd /your/gsdownload/dir # I got 8.50, the current gnu one.
> 
> 
> There are two parallel versions of ghostscript about whose relationship 
> I'm unclear. Version 8.50+ comes from www.ghostscript.com and is called 
> "AFPL Ghostscript". Fedora ships with the one from the GNU project, is 
> at version 8.16 and is obtained via ftp://ftp.gnu.org/ftp/ghostscript.
It _is_ confusing. I used 'GPL ghostscript' from:
http://www.cs.wisc.edu/~ghost/
version 8.50 as listed on that page. I haven't the foggiest idea how that 
number maps to gnu-ghostscript, CUPS-ghostscript, or anything else.
All I know is that I got mpl to shut up after installing this :)
Cheers,
f

Fernando Perez wrote:
> cd /your/gsdownload/dir # I got 8.50, the current gnu one.
There are two parallel versions of ghostscript about whose relationship 
I'm unclear. Version 8.50+ comes from www.ghostscript.com and is called 
"AFPL Ghostscript". Fedora ships with the one from the GNU project, is 
at version 8.16 and is obtained via ftp://ftp.gnu.org/ftp/ghostscript.
Steve

Hi,
I've been testing my matplotlib 0.86.2 installation on Fedora Core 4 and 
hit a problem I can't solve. If I do "python tex_demo.py" in the 
examples directory it seems to hang up. A quick "ps faxw" and a bit of 
digging shows that the problem is that on my installation of tetex, TeX 
isn't recognizing \rmfamily:
> freyer:~/.matplotlib/tex.cache> tex 0885e19fbc15d1bc2b5e1d3ac6018bee.tex
> This is TeX, Version 3.141592 (Web2C 7.5.4)
> (./0885e19fbc15d1bc2b5e1d3ac6018bee.tex
> ! Undefined control sequence.
> l.5 {\rmfamily
> \TeX\ is Number 
> $\displaystyle\sum_{n=1}^\infty\frac{-e^{i\pi...
>
> ? x
What is the problem here? I searched the mailing list and no one else 
seems to be having this trouble, so obviously it's my fault, but I can't 
quite figure out how.

Jim,
James Boyle wrote:
> The data I deal with regularly has 'missing' values. I am quite pleased 
> that plot handles this situation so gracefully.
> I tried a scatter plot with two arrays with missing data and was a 
> little surprised that it failed.
> Is this a bug or a feature?
I guess I would call it a missing feature. I added masked_array support 
to line drawing, but neglected scatter. I think I can do it later today.
Eric

The data I deal with regularly has 'missing' values. I am quite pleased 
that plot handles this situation so gracefully.
I tried a scatter plot with two arrays with missing data and was a 
little surprised that it failed.
Is this a bug or a feature?
I can appreciate that in the plot case one can plainly see the gaps 
where the missing data lie, whilst in the scatter case points would be 
silently eliminated. So one might be reluctant to have this aspect of 
the data hidden in the scatter.
I would like to have scatter handle missing data - i.e. if either array 
has a missing value then the point is not plotted.
I have looked a bit at the plot code and must confess that exactly how 
the missing stuff is implemented is not obvious to me.
--Jim

On Monday 23 January 2006 19:40, Christopher Barker wrote:
> Bill Dandreta wrote:
> > Most distros don't use gnu ghostscript but esp ghostscipt
> > <http://www.cups.org/ghostscript.php>instead, that's why is is version
> > 7.07.
I just fixed a bug in the dependency checking. I'm sorry this didnt get fixed 
before 0.86.2 was released, but I didnt know it was coming.
I just made esp ghostscript version 7.07 the default on my machine, and my 
first test with mpl was successful. If anyone has a script that will produce 
unacceptable results with ghostscript-7.07, please send it to me (and please 
note any changes you have made to default rc settings).
Thanks,
Darren

You can always get the latest canvas buffer from
figure.canvas.tostring_rgb(). A quick pylab example would be:
from pylab import *
import base64
plot(rand(100))
a =3D gca()
c =3D a.figure.canvas
buf =3D base64.encodestring(c.tostring_rgb())
# I would suggest using PIL
import Image
x =3D Image.fromstring('RGB', c.get_width_height(), c.tostring_rgb())
# You can then get specific format output strings
s =3D x.tostring('xbm')
On 1/24/06, Cyril Giraudon <cyr...@fr...> wrote:
> Hello,
>
> My problem is the following : I would like to create an image using the
> matplolib facilities and then insert this image
> in a <IMG SRC=3D"*data*:*image*/???;base64 *.../> *element.
> Then using Gecko (mozilla) and python + xpcom, I would try to interact
> with this image (zoom ...).
>
> In order to succeed I feel I have to understand how a back-end is written=
.
>
> It seems I have to start with the Figure class.
> But then I can't find the image format, the way to return binary data or
> perhaps base64 data.
> I don't want to save the figure in a file et then read it with <IMG
> SRC=3D"theFile.png"/>.
> The speed would be very low.
>
> What is the process to add a new back-end ?
>
> I wish the encode/decode base64 wil not take too much time for keeping
> the dynamism of gtk back-end (for instance).
> What do you think about that ?
>
> Thanks a lot,
>
> Cyril.
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through log fi=
les
> for problems? Stop! Download the new AJAX search engine that makes
> searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D103432&bid=3D230486&dat=
=3D121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Hello,
My problem is the following : I would like to create an image using the 
matplolib facilities and then insert this image
in a <IMG SRC="*data*:*image*/???;base64 *.../> *element.
Then using Gecko (mozilla) and python + xpcom, I would try to interact 
with this image (zoom ...).
In order to succeed I feel I have to understand how a back-end is written.
It seems I have to start with the Figure class.
But then I can't find the image format, the way to return binary data or 
perhaps base64 data.
I don't want to save the figure in a file et then read it with <IMG 
SRC="theFile.png"/>.
The speed would be very low.
What is the process to add a new back-end ?
I wish the encode/decode base64 wil not take too much time for keeping 
the dynamism of gtk back-end (for instance).
What do you think about that ?
Thanks a lot,
Cyril.

On Monday 23 January 2006 19:40, Christopher Barker wrote:
> Bill Dandreta wrote:
> > Most distros don't use gnu ghostscript but esp ghostscipt
> > <http://www.cups.org/ghostscript.php>instead, that's why is is version
> > 7.07.
According to http://www.cups.org/espgs/, the current version of esp 
ghostscript is 8.15.1
Darren

On 1/23/06, Darren Dale <dd...@co...> wrote:
>
>
> I have one question. If, for example, a user sets usetex =3DTrue in
> matplotlibrc, but TeX is not installed, is it better to issue a warning
> and
> force usetex=3DFalse, or is it better to raise an error? Right now I do t=
he
> former.
>
Issue a warning. At least the user gets a plot, even if it is not a good
one.
 -- Paul
--
Paul Barrett, PhD Johns Hopkins University
Assoc. Research Scientist Dept of Physics and Astronomy
Phone: 410-516-5190 Baltimore, MD 21218

Minor release to correspond with numpy-0.9.4 compatibility changes.
http://sourceforge.net/project/showfiles.php?group_id=3D80706
http://cheeseshop.python.org/pypi/matplotlib/
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
2006年01月24日 Released 0.86.2
2006年01月20日 Added a converters dict to pylab load to convert selected
 coloumns to float -- especially useful for files with date
 strings, uses a datestr2num converter - JDH
2006年01月20日 Added datestr2num to matplotlib dates to convert a string
 or sequence of strings to a matplotlib datenum
2006年01月18日 Added quadrilateral pcolormesh patch 1409190 by Alex Mont
 and Paul Kienzle -- this is *Agg only for now. See
 examples/quadmesh_demo.py - JDH
2006年01月18日 Added Jouni's boxplot patch - JDH
2006年01月18日 Added comma delimiter for pylab save - JDH
2006年01月12日 Added Ryan's legend patch - JDH
2006年1月12日 Fixed numpy / numeric to use .dtype.char to keep in SYNC
with numpy SVN

On 2006年1月23日, Andrew Straw apparently wrote: 
> http://new.scipy.org/Wiki/Cookbook/Matplotlib 
That's helpful.
A query:
has anyone implemented a nice SCATTERPLOT MATRIX
(see http://www.math.sfu.ca/~cschwarz/Stat-301/Handouts/node43.html
for an example or for variants see e.g.
http://www.ncss.com/scatmatrix.html 
http://www.itl.nist.gov/div898/software/dataplot/refman1/auxillar/scatplma.htm 
http://www.mrc-bsu.cam.ac.uk/bugs/documentation/coda03/node34.html )
Thank you,
Alan Isaac

Bill Dandreta wrote:
> Most distros don't use gnu ghostscript but esp ghostscipt 
> <http://www.cups.org/ghostscript.php>instead, that's why is is version 
> 7.07.
Which is a good reason NOT to replace it with Gnu Ghostscript, and that 
is a good reason to really try to fix MPL so that it doesn't require the 
latest Ghostscript.
But then, I'm not volunteering to try to figure all this out.
-Chris
-- 
Christopher Barker, Ph.D.
Oceanographer
 		
NOAA/OR&R/HAZMAT (206) 526-6959 voice
7600 Sand Point Way NE (206) 526-6329 fax
Seattle, WA 98115 (206) 526-6317 main reception
Chr...@no...

Hi Darren,
It may be very different than what you're doing, but the hybrid
approach of gnuplot's 'pslatex' terminal device might be worth
considering and looking into. This writes postscript for the graphics
part and leaves simple lines (for the axes) and the text as plain
latex that overlays the postscript. I think for mpl, you might want
to have all the non-text go into the postscript and all the text go
into an accompanying latex file, but the idea is the same.
One advantage there is portability, as the output is latex+embedded
postscript (using \special). It also allows pretty fine-grained
control on the output, including changing fonts or doing latex things
that mpl can't do (and this can all be done after the fact, so that
you can create the figure, and then change the fonts). It does
require latex to create the figure, but this step could be automated,
at least to stage of the dvi-with-eps-figure stage. Getting to ps,
pdf, or png would be less easy to automate but may be doable in a
portable way.
That sounds easier than parsing a dvi file to me. And postscript
backend already exists.
--Matt

= What's going on? =
The Matplotlib cookbook is hosted at scipy.org. The backend software 
that hosts the website is being changed (from Plone to MoinMoin). This 
morning I converted the matplotlib cookbook, which is hosted at 
scipy.org, for use with the new software. It is available at 
http://new.scipy.org/Wiki/Cookbook/Matplotlib . (This URL to the new 
cookbook will change once the transition is complete.)
Unfortunately in the transition period, there will be 2 copies of the 
cookbook available online:
 1) the shiny new cookbook at currently available at 
http://new.scipy.org/Wiki/Cookbook/Matplotlib
 2) the slow-to-edit and to-be-removed cookbook at 
http://www.scipy.org/wikis/topical_software/MatplotlibCookbook
*Please make any changes to the new cookbook!*
= Further information =
I've been spearheading an effort to move the scipy.org website to 
something more user-friendly. The goals were for the new site to be:
 * more inclusive of the scientific-computing-using-python community 
rather than the those-using-scipy-the-package community
 * easier (faster) for people to edit
 * more visually pleasing. (aka better marketing)
The decision we took was to make the new website based on MoinMoin, the 
wiki engine available at http://moinmoin.wikiwikiweb.de .
= What you can do =
Please review the Matplotlib Cookbook in its new location. If you can 
correct any errors yourself, that would be greatly appreciated. If you 
just drop me an email, that's fine, too -- I'll try to fix it. We'd also 
appreciate any help and suggestions for the entire new website.
Cheers!
Andrew

On 1/23/06, Christopher Fonnesbeck <ch...@tr...> wrote:
> On Jan 23, 2006, at 2:01 PM, Charlie Moad wrote:
>
> > "LDFLAGS=3D/tmp/static-libs python setup.py build".
> > This will make setuptools look in the directory you specify first.
>
> Thanks Charlie,
>
> Does that mean that it will static link to all libraries in /tmp/
> static-libs (for example)? Is there any problem with leaving mine in /
> usr/local/lib and specifying LDFLAGS=3D/usr/local/lib?
It doesn't matter where they are, but the linker will use whatever it
finds first. I think ld has a preference for shared libs, so if
libpng.so and libpng.a are in /usr/local/lib then they shared will
probably be used. Specifying "LDFLAGS=3D-L/usr/local/lib" (note my typo
before) ensures this directory is searched first.

On Monday 23 January 2006 12:34, you wrote:
> Darren Dale wrote:
> > The ghostscript requirement is something of a headache. The conservative
> > distributions, such as RHEL, continue to provide ghostscript version
> > 7.07
[...]
> Fedora Core4 is not generally a "conservative" distribution, and OS-X
> (10.3 anyway) doesn't provide 8.16 either. It would be great to not have
> this requirement.
I just spent some time improving the dependency checks, the changes are in 
cvs. I relaxed the ghostscript requirement to gs-7.07, but issue a warning if 
a version earlier than 8.16 is found. Based on Ryans post, and the recent 
changes to the mpl code, I will update the usetex wiki page sometime within 
the next couple days. 
I have one question. If, for example, a user sets usetex =True in 
matplotlibrc, but TeX is not installed, is it better to issue a warning and 
force usetex=False, or is it better to raise an error? Right now I do the 
former.
Darren

On Monday 23 January 2006 15:05, Ryan Krauss wrote:
> Hey Darren,
>
> Maybe this will help, maybe it won't.
> Ubuntu provides 3 gs options in its package manager:
> gs-afpl v. 8.14
> gs-gpl v. 8.01
> gs-esp v. 7.07
>
> I have also installed gpl 8.50 from source.
>
> My problems only occur when including the eps files in a latex file.
> With gs-gpl 8.01 or gs-esp 7.07, the output looks like the bad_bbox
> picture (it only shows the top of the figure where the plot is
> covering up the text). 
You can force extra space around the figure to prevent this overlap by doing 
something like this, I just dont remember which of the 4 values is relavant 
for this problem:
\includegraphics[trim=-10 -10 -10 -10]{rk-fig.eps}
> With gs-afpl 8.14 or gpl 8.50 (from source), I 
> get the output in better_bbox.jpg, but there is still a thin white
> line from the figure that goes through the text. 
This line can be removed by doing something like:
\includegraphics[clip=true]{rk-fig.eps}
> So maybe the white 
> frame or a white background is messing things up.
>
> But I don't think the problem really is the bounding box, I think it
> is this white frame or background being drawn bigger than the bounding
> box, because if I use the LaTeX command \includegraphics* (which as I
> understand it clips the figure to the bounding box - as opposed to
> \includegraphics without the * which does not), the output from all 4
> gs versions is fine. 
I wasn't aware of \includegraphics*...
> epstopdf makes nice pdfs with a good bounding 
> box/paper size regardless of which version generates the eps files
> (but I have to use the gatech-theis class for my thesis which does not
> work with pdflatex, so I need nice eps files).
>
> The only problem I had meeting the v >8.16 problem was getting gs to
> find my system fonts. This was solved by setting GS_LIB and exporting
> it in my .bashrc file. I found the font paths by running gs -h on one
> of the ubuntu provided versions (i.e /usr/bin/gs-gpl -h).
>
> Hope this helps.
>
> Let me know if you want me to test anything else. I really value this
> option and am glad to help with its development/testing/debugging.
Thank you Ryan, this is a valuable post.

Hey Darren,
Maybe this will help, maybe it won't.
Ubuntu provides 3 gs options in its package manager:
gs-afpl v. 8.14
gs-gpl v. 8.01
gs-esp v. 7.07
I have also installed gpl 8.50 from source.
My problems only occur when including the eps files in a latex file.=20
With gs-gpl 8.01 or gs-esp 7.07, the output looks like the bad_bbox
picture (it only shows the top of the figure where the plot is
covering up the text). With gs-afpl 8.14 or gpl 8.50 (from source), I
get the output in better_bbox.jpg, but there is still a thin white
line from the figure that goes through the text. So maybe the white
frame or a white background is messing things up.
But I don't think the problem really is the bounding box, I think it
is this white frame or background being drawn bigger than the bounding
box, because if I use the LaTeX command \includegraphics* (which as I
understand it clips the figure to the bounding box - as opposed to
\includegraphics without the * which does not), the output from all 4
gs versions is fine. epstopdf makes nice pdfs with a good bounding
box/paper size regardless of which version generates the eps files
(but I have to use the gatech-theis class for my thesis which does not
work with pdflatex, so I need nice eps files).
The only problem I had meeting the v >8.16 problem was getting gs to
find my system fonts. This was solved by setting GS_LIB and exporting
it in my .bashrc file. I found the font paths by running gs -h on one
of the ubuntu provided versions (i.e /usr/bin/gs-gpl -h).
Hope this helps.
Let me know if you want me to test anything else. I really value this
option and am glad to help with its development/testing/debugging.
Ryan
On 1/23/06, Christopher Barker <Chr...@no...> wrote:
> Darren Dale wrote:
> > The ghostscript requirement is something of a headache. The conservativ=
e
> > distributions, such as RHEL, continue to provide ghostscript version 7.=
07,
> > but we have had problems with the postscript output being clipped with =
this
> > version.
>
> Has anyone figured out what the problem is with older gs versions? gs
> has been around along time, there has got to be a way to generate EPS
> that would work with older versions.
>
> Fedora Core4 is not generally a "conservative" distribution, and OS-X
> (10.3 anyway) doesn't provide 8.16 either. It would be great to not have
> this requirement.
>
> -Chris
>
>
> --
> Christopher Barker, Ph.D.
> Oceanographer
>
> NOAA/OR&R/HAZMAT (206) 526-6959 voice
> 7600 Sand Point Way NE (206) 526-6329 fax
> Seattle, WA 98115 (206) 526-6317 main reception
>
> Chr...@no...
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through log fi=
les
> for problems? Stop! Download the new AJAX search engine that makes
> searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D103432&bid=3D230486&dat=
=3D121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Dan Christensen <jd...@uw...> writes:
> I'd like to plot a large number of points (say 10**6 or 10**7) with
> different colours. The colours will not naturally be parametrized by
> a single float, but rather the rgb components will all vary
> independently, so I don't think I can use a colormap.
...
> If I ignore the colors, and plot my data using one (or just a few)
> plot commands with marker '.', the speed is reasonable.
>
> 3) Scatter plots are close, but currently only allow a colormap
> indexed by a single float. I'm also concerned about efficiency,
> since every point seems to be plotted as a polygon, but I just 
> want small dots.
I have now verified that
 scatter(x, y, s=1, marker='o')
is quite a bit slower than
 plot(x, y, '.')
I tested with x and y both equal to arange(10**6).
The most noticeable difference is in (re)drawing the window, which
takes around 40s for the scatter plot and 1s for the ordinary plot.
It sounds like there are a few options:
1) Extend plot to allow an array of colour data.
2) Extend PolygonCollection to include points as a special case.
3) Make something like a PointCollection class which specializes
in drawing lots of points.
Can someone suggest which might be the easiest to pursue? And
which might give the fastest plots? 
Dan