[Axiom-developer] Re: Desired functionality from noweb

[C Y]

--- address@hidden wrote:

> Cliff,
> 
> Another concern is the ability to drag-and-drop a pamphlet
> onto axiom and have it "just work".

That will be heavily involved with the details of what platform, what
tools are needed by the pamphlet, etc.  Abilities that would be nice
for pamphlets but aren't supported by the existing tools are going to
be problematic.  (Compiler support for pamphlet locations in Lisp as a
replacement for the %def hack, for example.  I suppose in theory there
could be some kind of language independent mechanism implemented by all
compilers to store that information automatically in a universal
format...)

> We need to have an "adaptive" rather than "restrictive" approach,
> I think. That is, we need to be able to adapt to xy vs xypic
> as long as it doesn't break things. We need to be able to handle
> single vs multiple pamphlets. 

Perhaps I should take a step back and make some "use cases" of things
so I have a better sense of how this is supposed to work.  I have a
feeling I might be making some fairly fundamental assumptions that are
not particularly valid.

> All of this can evolve with time, as the requirements arise.
> 
> But it will be most helpful if we can rise up out of the dust
> of particular syntax and define the algorithms in terms of
> functions as I mentioned earlier. Then the algorithms can be
> expressed as graph-to-graph transforms and the low level details
> can be handled by the pamphlet->graph and graph->pamphlet functions.

That sounds good but I'll need to think about it some more to
understand what it really implies.

> Graph-to-graph transformations occur naturally in lisp.
> Any graph can be easily expressed in lisp list notation
> so any algorithm is just a manipulation of that list.
> 
> Better still the lisp list graph can executed as well as manipulated.
> 
> Thus, given the latex-to-lisp tranformation
> 
>   \section -> (SECTION
> 
> and a definition of a SECTION function means that the graph
> can have both structural and execution semantics.

Sorry for being dense, but can you illustrate what the possible
consequences of that would be?

Cheers,
CY

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com