Re: [Axiom-developer] literate programming, Claerbout's Insight and Doyen

[Ralf Hemmecke]

On 07/29/2007 06:29 AM, address@hidden wrote:
> Now that memory kicks in I recall that one original motivation 
> for the Doyen project was to create particular "snapshots" for
> a particular piece of research that could always be reproduced.
>
> Jose has carefully documented the process of creating a Doyen
> CD so a paper could be published on the CD along with a complete
> Axiom system to sit behind the paper.
>
> The idea is that you can always reproduce and rerun the paper
> even if the world moves. So Axiom development could continue
> but you could still run last year's papers.
>
> Thus Martin could build and distribute an environment that has
> the necessary scripts to run his algebra. Or Ralf could distribute
> the combinat environment on a Doyen CD along with traditional
> publications.

CDs are, of course dying out. I would rather like to follow the idea of
LogiWeb. (Recent thread starts at 
http://lists.nongnu.org/archive/html/axiom-developer/2007-06/msg00658.html)

LogiWeb already has the idea of being like a journal. It's like a 
version control system but about scientific papers. Now, if those 
scientific papers contain code or not does not matter. If they contain 
code, all the better. Starting from the references of a paper, it can be 
found out on which other papers (code) the paper rests. So the paper 
plus the logiweb infrastructure actually gives a way to produce exactly 
the same running system as the author had when he produced the paper.

Of course there has to be some thought of how bugs are fixed in such a 
system and how one deals with progressing underlying hardware. But the 
basic idea should be like that.

For bugs, I think one has to review the "change file" idea of Knuth. (I 
did not find time to even find a good reference. But time will come when 
that will be important.)

Ralf