- From: William F. Hammond <hammond@csc.albany.edu>
- Date: Tue, 11 Apr 2000 20:16:36 -0400 (EDT)
- To: hutch@psfc.mit.edu
- Cc: www-math@w3.org
Ian Hutchinson writes:
> content. TeX-the-programming-language can be programmed to do most
> things. The question is, though, whether it is a good choice of
> language for the purpose in mind.
I didn't use to think so. But as the Perl folk say, TMTOWTDI. :-)
> I would argue that since expressing Content
> requires you to change completely the semantics of TeX (relative to
> Plain or LaTeX for example) in writing equations, it would be an
> arbitrary choice to use it.
Or else think *very hard* about how the human reader is able to
correctly parse printed output, and then, understanding that, figure
out how to add what the human reader understands implicitly.
Bear in mind that authors will be impatient with systems that are
inefficient for them.
> The point is that you would require a "package" that completely
> changed the whole manner of entering equations. Again, yes that could
> be expressed using TeX-programming, since the syntax of TeX can be
> changed almost arbitrarily, but it would be unrecognizable to authors
> as the TeX they know and (like us) love.
Simplified example:
In preamble:
\declaremathsymbol{D}{operator}
\declaremathsymbol{y}{function}
Armed with that information along with rudimentary knowledge at the
level of *relative semantics* about "operator" and "function" (such as
[oversimplified] "functions act on numbers", and "operators act on
functions", and the sensible rule that operator multiplication is
composition), I expect that most robots can be easily taught how to
parse $D^2 y$ as content markup.
And my hunch is that less will be more as long as less is enough.
-- Bill
Received on Tuesday, 11 April 2000 20:17:12 UTC