Print

Print


At 16:33 97/06/26, Sebastian Rahtz wrote:
>while it is true that saving the cost of capturing the `formula' is a
>good thing, it is often outweighed by the cost of dealing with the
>rest of the abysmally LaTeXed article. I am, I confess, getting very
>tired of having to defend the conventional publishers like my
>employers, but you must realize that rekeying of material by quite
>low-level workers in the Philippines is surprisingly often more cost
>effective than dealing with the LaTeX submission.

  Rekeying of any material is not cost-effective in the long run
developments perspective, even though it may be a workable cost-effective
solution in the short term, which will be more apparent as the information
flow increases in the future. Also, one problem with rekeying formulas, is
that they seldomly come out right, and the correction process is not
convergent; one idea with TeX was to avoid that.

  Another problem is the lack of flexibility with LaTeX, which is
well-known, and the LaTeX3 project should aim to amend that. For example,
the original LaTeX style goes against what one would expect in a math paper
in many respects, and it is not easy, in the way it is easy in MSWord or
some WYSIWYG program, to change that. When you write something, you do not
want to be a programmer, because it too distracting.

  For simple formulas, one might use WYSIWYG equation processors, but for
complex formulas, this does not work well, if one cannot somehow use
macros. Another problem with WYSIWYG formulas is that these are usually
graphical entities, so they are difficult to reuse. So I think LaTeX3
should promote a development towards entering more structured formulas
(this has been discussed in the math-fonts group). But I think this would
need a new TeX: Having a enhanced TeX which could also parse formulas (like
in a user definable compiler-compiler), and select typesetting style based
on that, would surely help the author entering formulas.

> >   So the ideal for the author, is probably to quickly output the semantic
> > contents, and it should be possible to somehow enter the other stuff
> > independently of this (also so that it is possible to choose a
> > suitable
>
>so, you enter the semantic content in SGML, and twitch a DSSSL style
>sheet which describes the layout, to your hearts content

  I am not sure what bearing this has on the LaTeX3 project. The best way
to go, would perhaps be to let LaTeX3 better split up the material in
authoring/style, so that those who typesets it, with whatever method they
feel happy with, can more easily extract just that information they need.

  (I had the idea of a LaTeX "TeX" documentclass, which would make LaTeX
behave as regular TeX. Users importing TeX material to LaTeX could use this
class for a start, and then add other LaTeX-features to this.)

  Hans Aberg