[SciPy-dev] old web site front page
jh at oobleck.astro.cornell.edu
Wed Dec 21 13:42:09 CST 2005
> I think we should be more enthusiastic about recommending that people
> migrate now to the new SciPy.
Ed, I agree that there are people we should be encouraging to try the
beta: current, experienced SciPy users on platforms with binary, full,
newcore builds. Currently there are no full builds on any of the
public servers. There is also no documentation (Travis's book is
still a pre-order), and as we've been discussing the web sites need
serious work. Even the most far-along builds are still beta for these
reasons. I think encouraging anyone who is not already a user or
potential developer would be a mistake, because we'd risk losing them.
There are a lot of such people, and the main-page text has to address
everyone. I've incorporated some of your ideas in the revised text,
below. This is now the full text I propose for the page.
> I've added the first paragraph of the text proposed in the previous
> message to the front page. And the rest to installation instructions.
> I didn't know where to put the plea for voulnteers.
I'm proposing this text for the current (Plone) site, since that's
where newbies go. No problem cribbing it for the new site, but
remember that when the new site goes live our recommendations will
(hopefully!) be different from what they are today.
WHAT IS SCIPY?
SciPy is Open Source software for mathematics, science, and
engineering. It runs on all popular operating systems, is quick to
install, and is free of charge. SciPy is easy enough that a
12-year-old can use it to plot up her math homework, but complete
enough that it is depended upon by some of the world's forefront
scientists and engineers. If you need to manipulate numbers on a
computer and display the results, we hope you'll give SciPy a try!
SciPy consists of two packages. The "SciPy Core" provides basic
functionality for array mathematics. "Full SciPy" provides modules
for statistics, optimization, integration, linear algebra, Fourier
transforms, signal and image processing, genetic algorithms, ODE
solvers, special functions, and more. In addition, well over 100
application software packages use SciPy for specific tasks, and the
number is growing.
SciPy is nearing completion of a major overhaul. The new SciPy
features high performance for both small and large arrays, and rich
syntactic functionality for both. This effort is mostly done. We are
now beating the bugs out of the various platform builds to make stable
binary installs. Documentation is also well underway, as is a web
WHAT TO DO NOW:
If you are a CURRENT SCIPY USER, we encourage you to give the new
version a try. Source code and binary installers for the latest
beta-test version (for a few platforms) are available from
http://numeric.scipy.org. If a binary package is available for your
computer, things should be relatively stable.
If you are NEW TO SCIPY, there are several reasonable paths to take,
depending on your circumstances:
1. If you can wait until the Spring of 2006, the new package, binary
installers, and its basic documentation should be ready by then. This
is the path of least resistance!
2. If you want to try the new SciPy anyway, you can try the beta
above. Look to the old SciPy documentation, the tutorials for the
Numeric and Numarray packages, and the scipy-user at scipy.net mailing
list for help.
3. If you must run an application that requires either old version of
the array extension to Python (Numeric or Numarray), go ahead and
install that version. The latest versions of both Numeric and
Numarray support importing and exporting SciPy Core arrays. They will
both be viable for some time to come.
In any case, if you are VERY EXPERIENCED WITH SOFTWARE INSTALLS and
like getting your hands dirty, download the tarball for the new
version, build it for your architecture, take notes, subscribe to
scipy-dev at scipy.net, and post your notes once you've read recent
threads in a similar vein. Your experience will help us, and other
new users, a great deal. In addition to the official download site,
see http://new.scipy.org/Wiki/Download for SVN access to a version of
full SciPy that works on newcore.
The official release of the new SciPy is due in Spring 2006. It will
be similar to the beta, but with more documentation, binaries for more
platforms, and more bug fixes.
Old (Numeric-based) SciPy is available from
http://www.scipy.org/download/ and has a Yum repo at
http://www.enthought.com/python/fedora/. Most of the material on this
web site refers to the old, Numeric-based SciPy.
We strongly recommend subscribing to both scipy-dev at scipy.net and
scipy-user at scipy.net while the software is still in testing, and
updating whenever a new release occurs.
WE NEED VOLUNTEERS!
At this point, if you just do things for yourself and post what you
do, you'll make a big contribution. Initially, that will help us get
the code into release-stable form. Eventually, it will make a
friendly and easy set of installation instructions a reality.
We are also looking for people to write documentation and to help
design and populate the new SciPy web site. The new site is
SciPy is a community project that is sponsored and supported by
More information about the Scipy-dev