[Axiom-developer] Re: Axiom UI

[Marc Battyani]

On Thursday, July 07, 2005 10:07 PM Bill Page wrote:

> On Thursday, July 07, 2005 3:59 AM Marc Battyani wrote:
>
> > I've looked at the Axiom UI SOC project and I thought there may be
> > some possibility to use cl-typesetting/cl-pdf as a typesetting engine
> > for the layout of the mathematical expressions.
>
> Thanks for you email. Yes, we have already briefly discussed
cl-typesetting
> as an option for the new Axiom user interface.
>
> The major part of the proposal by Kai Kaminski which was selected for
> the SOC funding concerns a web browser based interface. See:
>
> http://page.axiom-developer.org/zope/mathaction/AxiomUI

OK. I didn't know that. (it's not mentioned on
http://www.alphageeksinc.com/cgi-bin/lispnyc.cgi?AxiomUI)

> However, that does not preclude the possibility of using tools like
> cl-tempsetting, especially in the context of high quality printable
> output such as PDF, as an alternative to more traditional routes like
> LaTeX.

Yes, but int the case of a pure web interface I agree it's better to use
jsMath and SVG. The math stuff of cl-typesetting is not finished as nobody
has really used it. I mainly added some math capabilities more as a proof of
concept than for a real need.

> > cl-typesetting already has math capabilities and can output to a pdf
> > file but it is also possible to write to a screen or any graphical
> > device. It is already platform independant and of TeX like quality.
>
> Can you point us to any example programs that run on both linux
> and windows? What flavor of lisp provides the greatest portability?

I use it both on linux and windows in web applications to generate all kind
of documents, like documentation, business reports, traceability reports,
calibration certificates, physical data measurement (plots), etc.
The implantation I use is LispWorks but cl-pdf/cl-typesetting is known to
work on opensource implementations like CMUCL, SBCL and CLISP. (well may be
the latest SBCL don't work due to the changes in external format)

> > I can even help on that if needed and add stuff such as finding which
> > element got a mouse click (for editing purposes) for instance.
>
> I would like to hear more about how you see this working. In particular,
> what would be the format of the expressions which could be edited in
> this manner? How closely could then correspond to the mathematical
> content (as opposed to the presentation) of the output of the computer
> algebra system?

It's just an idea. I think that from the click place it's possible to go
back to corresponding text/form in the formula input. But you can do that in
Javascript as well in jsMath. (just put an on_click handler in the span
elements)

> > In addition to this intereactive mode, a cl-typesetting/cl-pdf backend
> > could be useful too.
>
> I agree.
>
> > For an example of math layout, look at page 3 of :
> > http://fractalconcept.com/fcweb/download/ex.pdf
> >
> > The lisp to generate this formula is:
> > (display-formula ()
> >        (with-style (:font (pdf:get-font "CMEX10" nil) :font-size 30)
> >          (with-offset (23) "H"))
> >        "E"(math-super-and-sub-script () ("n+1") ("k,m"))"="
> >        (fraction ()
> >    ("x"(with-superscript () "2")"+x-1")
> >    ("F(x)+b-3"))
> >        "-e"(with-superscript () "-x"(with-superscript () "2")))
>
> Have you considered what might be required to convert from, say
> LaTeX equations syntax to this sort of lisp? Or from a (much) simpler
> s-expression format such as might be output by a computer algebra
> system like Axiom?

I didn't look at the LaTeX syntax in details but using a simpler
s-expression format would be easy.

> Thanks for your interest. I think we need to be aware of these sort
> of advances and also to proceed on more than one front -- if resource
> would permit it. Unfortunately resources are in ratjer sort supply...
> :) But if you have an interest in Axiom and want to proceed with some
> kind of demonstration or whatever, I am sure you will have the ears
> and eyes of other axiom developers.

Well, I really think that given the choice of a browser as the UI engine
(which is a very good choice IMO) it's better to stick to native browser
stuff like Javascript/SVG/jsMath.

Marc