[MATRIX-SIG] Revised proposal for PLOT-SIG

[Paul Barrett]

I've been following along on this thread for awhile now, but have been
too busy doing programmatic work to take the time to respond.  I'll
make some time and do so now.

Some of my current feelings are:

1) A PLOT-SIG should be created in spite of what happens to the
MATRIX-SIG.  The creation of a SIG with narrow focus and specific
goals is more likely to reach its objective than being a splinter
group in a more general SIG like a DATA-ANALSYS-ENVIRONMENTS-SIG.
Note that the MATRIX-SIG was created for just this reason.

(As an aside, I don't think the MATRIX-SIG should be terminated, but
just moved to the back burner, so that the old-pros and newcomers can
revisit this topic from time to time without creating noise in newly
created or related SIGS.)

2) Now that the MATRIX-SIG is nearly complete, a PLOT-SIG (or
GRAPHICS-SIG, or DATA-DISPLAY-SIG) is of next highest priority for me.
Where I work, IDL and SuperMongo are still used by my coworkers
because of their plotting capabilities, mostly for their ease of use.
Note that they both have their limitations, mostly associated with
their inferior language and syntax and user interface.  Many coworkers
would be willing to migrate to Numeric Python if there was good
plotting capabilities, eg. 2-D, contour plots, shaded-surfaces, etc.
An excellent plotting package able to do 3-D volume rendering and
animation alla Vis5D could be developed later, but should be included
in the initial design.

3) I also feel that we should try to develop a high level OO interface
to current packages and if deemed necessary a new Python developed
package.  The former would help to get some good graphics capability
into Python quickly using pgplot, PLplot, etc., but would allow
migration to a more advanced plotting package in future.  I don't see
why those of us wanting to continue to use one of the currrent
packages cannot add any addition functions to get full capabilities
from their package, knowing full well that these functions are not
platform and package independent.

4) My current dismay at many current packages is that they are
platform specific and do not really allow easy access to low level
features of the graph.  I see a design with 3 or 4 levels.  The
highest level would provide the creation of a simple plot from a
single command.  Lower level commands would allow users to create
new standard plot types (eg. bar charts, etc.) using the low level
commands. 

In addition a high level interface would allow us to develop a
platform independent package.  IDL has managed to do this across X11,
Windows, and Mac interfaces.  I don't see why we can't do something
similar.

6) I think we should create the SIG and start hashing out the design
issues ASAP.  A table containing a consensus of each plotting packages
strengths and weaknesses might help immensely in our final design.

 -- Paul


_______________
MATRIX-SIG  - SIG on Matrix Math for Python

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