[Axiom-developer] (no subject)

[daly]

Alex, Heow,

Axiom is implemented mostly in Common Lisp. The original system had
code which was in C (hyperdoc and graphics). It also had 2 layers of
pre-lisp languages, Meta and Boot. 

Meta was used to specify syntax. I rewrote that code and Meta is no
longer part of the system, simplifying the build process.

Boot is "syntactic sugar" on top of lisp. It is a "paren-less lisp"
with indentation-based syntax like python. Boot is implemented in
boot so there is a bootstrap step (hence the name). I am in the
process of removing this pseudo-language as it:
 * unnecessarily complicates the build (bootstrap)
 * foils the use of lisp tools like macros
 * does not support structures, clos, and other machinery
 * perpetuates MACLISP and VMLisp historical semantics
 * is an obscure language with no known users  

In the near term (probably by the end of the year) Axiom will be
all common lisp, except for hyperdoc and graphics. 

Hyperdoc and graphics are being reimplemented using a firefox front
end and javascript. Most of the static content of hyperdoc is now in
xhtml. The remaining effort involves writing ajax pages for dynamic
help content and canvas-based graphics. Both are in-plan. So by
Christmas 2010 Axiom will be a common lisp program with a firefox
front end in javascript/xhtml/mathml in fully literate structure.

Following that goal, Axiom needs to recover the numeric library
capabilities. The NAG libraries are not free so they need to be
replaced. I have been rewriting the BLAS, LAPACK, and GMP sources into
literate form and will eventually add an additional numeric volume to
the Axiom bookshelf.




The time for a transition to literate programming is perfect from an
historical perspective because we are beginning to see a trend toward
open and free research literature. MIT, Harvard, NIH, and other agencies
are beginning to understand that science needs to be free. This means
that we can finally bring the research and the implementations back
together in one document.



If we can create fully literate documents that contain the research
and the implementation then we can begin to think about "drag and drop"
mathematics. You should be able to attend a talk where the speaker
posts the URL for the paper. You should be able to drag that paper
onto a running Axiom and have it "just work", including installing and
using the documentation. Thus, while the talk is in progress you should
be able to use the software to execute examples. See:

  http://daly.axiom-developer.org/doyen

Tim