Re: [Axiom-developer] Re: literate programming pamphlet filesforMathAction

[Mike Dewar]

On Sun, Oct 03, 2004 at 10:20:54AM -0400, William Sit wrote:
<snip>
> I also think that hyperdoc has a lot more functionality than TechExplorer
> offers. It is true that hyperdoc does not give pretty font output (especially
> math), but I'll pass over beauty for utility anytime. Formatting is a very
> painstaking job and difficult to do it right (even when done manually and
> interactively). It is great that you and so many others devoted so much time 
> to
> improve Open Axiom.

Actually there isn't a single thing that you could do in HyperDoc that
you couldn't do in our TechExplorer-based Windows interface, and in
practice the latter was much more natural to work with because it was
more closely integrated into the standard Axiom user-interface.

To clarify, there are two parts of HyperDoc.  The first is a set of
static pages (such as the Axiom book) which are indexed in different
ways.  In some cases links are embedded to allow an Axiom interpreter
frame to be spawned and commands executed in it (either with fixed data
or with data entered via a form-style interface) but that's really the
limit of the dynamic behaviour.

The second is the browser which is driven from the Axiom databases and
generates pages dynamically.  This means that if you add your own code
or change existing code in an Axiom session then the browser view will
reflect this.  You can inspect the exports of a particular type, browse
through its inheritance structure, find all domains which belong to a
particular category ...  Quite honestly I'm amazed (and impressed!) that
anybody can use Axiom seriously without the browser!

When we moved to CCL and ported Axiom to Windows we had to completely
re-write the browser code because it was entirely unix-specific.  This
coincided with a re-design of the databases to make them more compact
and faster to access (a lot of this work was actually done by IBM, and
particularly by the late Dick Jenks).  Unfortunately part of that work
included adding a "grep" function to CCL, so unless that code is ported
to GCL you'll have problems getting the existing browser to work in
OpenAxiom.

As far as user interfaces go, under Unix we stuck with the X11-based
HyperDoc, but under Windows we used TechExplorer.  The static pages
described above work without Axiom being present in both cases (although
obviously the dynamic bits won't work).  The browser requires Axiom to
be present, and on Unix this implies that sman is available.

Personally I'd have thought that you could adapt the TechExplorer
browser code to work with either the TeXmacs or Mathaction interfaces
since in principle they're all very similar.  You'll need to solve the
"grepping with GCL" issue first though.

Cheers, Mike.

________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________