Re: Steps towards writing a computer algebra system

From: Allan Adler (ara_at_nestle.csail.mit.edu)
Date: 01/30/05 

Date: 30 Jan 2005 01:00:41 -0500

What about *reading* a computer algebra system? I believe that it is
good to have a computer algebra system that comes with complete source
code, but as a practical matter, I find I am usually unable to read
the source code. It's usually more complicated than that, since the
source code for the computer algebra system is often written in a
substrate which itself, hopefully, comes with complete source code
in terms of a standard programming language such as C. I suppose Lisp
is considered a standard programming language, but there are various
dialects of Lisp and I tend to get lost sorting them all out, so having
a Lisp-like substrate tends to put me in the position of needing to
study the source code for the substrate. At that point, I give up
because I don't have enough time or ability or enough time to acquire
enough ability.

Anyway, I think it would be a great contribution if someone were to take
one of the major free computer algebra systems that come with complete
source code and explain in detail how it is written from the ground up
and how its various capabilities are activated and in what order when
the CAS is booted up. If that requires explaining the same thing for
the substrate, so be it.

If Knuth could do that for TeX and METAFONT, both their usage and their
underlying programs, in his computers and typesetting series, why can't
someone do it for a major CAS? Well, one reason is that Knuth froze TeX
and METAFONT at a certain level of perfection, while other CAS always
seem to be changing. But some of them are certainly good enough that
some version can be arbitrarily frozen for the purposes of dissection.

I've had a certain amount of experience taking computer programs that were
not written in a version of WEB or CWEB or FWEB or whatever and trying to
convert them to those formats as a way of studying the source code. It was
my bright idea on how to understand the source code and it was indeed
instructive but always ended in failure after a lot of effort. If successful,
the result would be something along the lines of what Knuth did, but precisely
the disability in reading a CAS is the main obstacle to writing the WEB
version. 

-- 
Ignorantly,
Allan Adler <ara@zurich.csail.mit.edu>
* Disclaimer: I am a guest and *not* a member of the MIT CSAIL. My actions and
* comments do not reflect in any way on MIT. Also, I am nowhere near Boston.

Relevant Pages

  • Re: GPL vs LGPL vs CAS
    ... >> perspective of CAS would not add much to the discussion. ... > to have access to the source code of a CAS. ... year writing a faster version of an algorithm, ... If Wolfram gave away Mathematica, ...
    (sci.math.symbolic)
  • Re: GPL vs LGPL vs CAS
    ... CAS or numerical computation software. ... to have access to the source code of a CAS. ... the source code of the algorithms in a CAS ... > Because the US funding agencies do not see system building ...
    (sci.math.symbolic)
  • Re: 49G+ CAS oddities, help!
    ... | Contains the source code for the CAS in the HP 49G calculator. ... You can recompile only the CAS parts of the ROM; ...
    (comp.sys.hp48)
  • Re: Steps towards writing a computer algebra system
    ... I have put together a minimal CAS which you may extend. ... interpretation of expressions and construction of a tree structure ... > Looking at source code to currently available open source pacakges is ...
    (sci.math.symbolic)