[PYTHON MATRIX-SIG] Plotting packages

Paul F. Dubois dubois1@llnl.gov
Sun, 9 Mar 1997 09:14:39 -0800

Some remarks about activity at LLNL:

We have had an OO attitude from the first about how the Python/graphics
interfaces should look, but also recognize there has to be an easy 

plot(y,x, color="red", style="dashed", scale="linlog") 

sort of interface for casual use. It is possible to do both, and we are doing

We have tried to create our work in layers so that the graphics object and
plot(y,x, ...) sort of layers are INDEPENDENT of the underlying graphics
library choice. It could be a reasonable goal for us to try to agree on the
design of the upper layers first, since the (time-dependent) question of
portability obfuscates the discussion of the bottom layers. Our design is
influenced (presumably positively) by our years of Basis and Yorick
experience, but since we do not all do the same kind of work I would presume
there are needs that our design does not meet.

On the subject of graphics packages, there are graphics packages and then
there are graphics packages. In this discussion little has been said about
the capabilities of the packages mentioned. It is one thing to need to graph
a curve; another to need to do colored contouring on irregular grids of data
with missing values. The set of available output-file formats matter, too.

The Gist module we have under development is very full featured. As Zane
mentioned, It is taking more time than we would like because we are both
underfunded and having to do a 3-D layer because of our application, but we
should have a much enhanced release for "PyGist" soon. The comment has been
made that Gist uses X-Windows only. I believe one of our physicists has
ported the C Gist library to NT and Mac. I am checking on this and will
report back. We are investigating how hard a Tk widget for Gist would be.

PLPLOT has been interfaced to Python with great success. It is also a very
capable library and a Tk widget for it has been developed. While there have
been isolated reports of difficulty installing it, the majority of people
seem to have no trouble. It would be helpful for people having difficulty
with an install to send details to furnish@llnl.gov. 

We just learned last week that a group doing climate modeling here is
beginning to interface their VERY fancy graphics library and accompanying
tools to Python. Python is being used as the "glue" to connect modeling
codes, data, and graphics. This library was written here and is entirely
self-contained. Again, as I learn more about it I will try to keep you all
informed.  This is presently a Unix library but the author said NT was on his
mind these days.

We consider portability to NT a longer-term necessity too. 

-- Paul Dubois

> From: Jeffrey Templon <templon@studbolt.physast.uga.edu>
> To: Jim Hugunin <hugunin@mit.edu>
> Cc: MATRIX Sig <matrix-sig@python.org>
> Subject: [PYTHON MATRIX-SIG] Plotting packages
> Date: Friday, March 07, 1997 12:49 PM
> Hi,
> Jim Hugunin writes:
>  > Recent posts have started me thinking about plotting and NumPy again. 
>  > seems to be the weakest link in the package at the moment (since I've
>  > been receiving minimal bug reports for the first beta release).  I'd
>  > like people to be able to type:
>  > 
>  > plot(sin(arange(0,5*pi,pi/10)))
> I'd like to reiterate my opinion that it'd be nice to have a plot be
> an object.  This allows us to move up a notch (out of spaghetti-like
> code) and do cool things like modify existing plots:
> 	plot1.xlims(10,20)
> 	plot1.update()
> to change the x-axis limits, or for example
> 	gwin.stackplots(2,2,plot1,plot2,plot3,plot4)
> to arrange four plots in a 2x2 matrix on the graphics window 'gwin'.
>  > Am I missing any packages?  Am I wrong in my comments about any of
>  >  It sounds to me like the Astronomers among us are only interested in 
>  > PGPlot.  Is this for historical reasons, or are there some features in
>  > package that are not supported by the others?
> Have you looked at DISLIN?  It's available in C, and looks pretty
> nice.  The Linux and DOS versions are free, but unfortunately you have
> to pay for the versions for other operating systems.  That's the
> biggest disadvantage.  Possibly you might cut a deal with them as part
> of making it a package.
> The URL
> 	http://www.mpae.gwdg.de/dislin/dislin.html
> 					JAT
> _______________
> MATRIX-SIG  - SIG on Matrix Math for Python
> send messages to: matrix-sig@python.org
> administrivia to: matrix-sig-request@python.org
> _______________

MATRIX-SIG  - SIG on Matrix Math for Python

send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org