RE: [Axiom-developer] [TeXmacs]

[Bill Page]

On Saturday, December 18, 2004 7:22 PM address@hidden
wrote:
> 
> I tried also with the same input but with power -1/3 instead 
> of 1/3,

  I1 := integrate (1/(x**3 * (a+b*x)**(-1/3)),x)

> and the result appears to have big left and right parentheses,
> but the first two lines are incorrect.

Actually if you look closely, it is correct but just formatted
in a rather unexpected style. To see this try

  numer(I1)

You will notice that in the full expression, the first term of
the numerator has been folded locally at the \cdot into two lines
with a leading minus sign vertically centered on the right like
this

   b^2x^2\sqrt{3} \cdot
-                               + 2b^2x^2\sqrt{3}log(sqrt[3]{-a^2} ...
   \log(sqrt[3]{-a^2}^2 ...

This type of "nested folding" is actually built in to Robert
Sutor's algorithm. There may be an option that I am not explicitly
setting which controls this type of behaviour.

See also the web page at

http://page.axiom-developer.org/zope/mathaction/VeryLongLaTeX

I am glad to see that all of these examples work on MathAction
since MathAction is actually using LaTeX internally to do the
formatting. So this represents a reasonable "benchmark" against
which to compare the TeXmacs output.

> So the error seems not to be localized to the treatment of 
> \left( and \right).

Yes, I agree. This is true in at least the current version of
TeXmacs. I seem to recall somewhat different behaviour in the
versions of TeXmacs that were current at this time last year.
The current problems might have something to do with how TeXmacs
converts \begin{array} structures into tables and an interaction
of the table entries (where each table entry is a folded line)
with complex math formatting.

> 
> On the other hand, the output of the cygwin tm_axiom is 
> correct, although on a single line. So I think that the
> only solution for users, for the Windows version of Axiom,
> would be to include the cygwin tm_axiom, until Andrey or
> Joris provide a corrected import module of LaTeX arrays in
> TeXmacs ?

So far I only know one example that fails to display properly.
Unfortunately it is the very first example in the book! Arghh.
One example of an apparently wrong result that is not easily
distinguished as a typesetting problem is enough to "stop the
show", of course. However I am optimistic that this can be
corrected quite easily since other apparently equally complex
examples actually do work.

I think that the more we (Axiom developers) depend on TeXmacs,
the more we must be prepared to try to help maintain TeXmacs -
just as we already have a great (sometimes two-way) relationship
with the GCL developers, on which the current version of Axiom
strongly depends ... :) That is just the nature of complex open
source development projects, I guess. So, as time permits, I hope
to be able to begin to work directly with the TeXmacs source code.
But so far it looks as foreign to me as Axiom's "bootsys" level
did to be two years ago ... <sigh>.

> 
> And congratulations for the great work !
> 

Thank you for your interest in Axiom in TeXmacs and agreeing
to test this!

Cheers,
Bill Page.