On Friday 02 February 2007, Jaap Spies wrote:
> Valery Pipin wrote:
> > Jaap Spies wrote:
> >> http://modular.math.washington.edu/sage/
> >> http://modular.math.washington.edu/sage/components.html
> >
> > I'd love to try it if all those components were distributed separately.
> > There are many reason for this
> > 1)84Mb ? source code and your site is slow
>
> Not so much to have all batteries included. Try a mirror site, e.g.
The problem is that I have nearly all of this in my distro(altlinux). The
similar can be said about debian, fedora, gentoo and others.
> http://sage.scipy.org/sage/
>
> > 2)The particular user need only a few componets, e.g. , I would use
> > python, numpy and maxima, mathplotlib.
>
> All the other goodies are there. If you don' need them, don' use them.
> It is free software.
>
> > 3) Most linux distribution have python, numpy, maxima and etc.
>
> From a system integrator's standpoint a possible nightmare:
> different versions, non-standard implementations. Some
> systems are extremely difficult to build without glitches.
Consider the rpm or deb based distro.
The real nightmare can start after installation.
What about system libraries zlib,libreadline, bzip?
Without these packages integrity of the system will be destroit!
>
> > 4) Sage use clisp to compile maxima (at least it was so 1 year ago). This
> > is really slowest choice. I admit that many other people can prefer clisp
> > over sbcl and cmucl. But it is always better to have a choice in this
> > option because maxima gives this choice.
> > ....
>
> I do not know much about lisp in all it's variations, but maxima in SAGE
> builds on clisp on a variety of architectures.
Is the computational efficience also a purpose?
Cmucl, sbcl and gcl exist on a variety of architectures as well.
In the current state it seems reasonable to distribute SAGE on the live cd as
well, e.g., as a knoppix subset.
best luck,
Valery