[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

More on subscripts and superscripts

As far as arrow control sequences in sub- and super-scripts goes, I
think we have three options:

1) Allow all of the CM* and MS*M arrows to be accessed directly by
\mathchardef's.

2) Allow the CM* arrows to be accessed directly by \mathchardef's, but
allow some of the MS*M glyphs to be accessed indirectly (e.g. by
ligaturing and kerning).

3) Allow none of the arrows to be accessed directly by \mathchardef's.

Advantages of each approach:

1) All (La)TeX and AmS(La)TeX documents will be compatible.

2) All (La)TeX documents will be compatible, although AmS(La)TeX
documents which include input such as `$X_\nrightarrow$' will break.
Most of the arrow glyphs will be accessed indirectly via the
arrow-building kit, which cuts down on the number of glyphs required.
All of the directly-accessed glyphs are very common, so should be
available in every math font.

3) The arrow-building kit will be simpler and smaller, and won't have
nasty kluges to allow for directly accessed glyphs.

My preference is to go for option (2), as a reasonable compromise
between compatibility and cleanness.  This means that documents which
contain input like `$X_\nrightarrow$' will break.

The TeXbook states that a subscript is of the form:

   <subscript><math field>

where:

   <math field> ::= <math symbol>
                  | <filler> {<math mode material>}

   <math symbol> ::= \mathchar<15-bit number>
                   | <mathchardef token>
                   | \delimiter<27-bit number>

So it's kosher plain TeX to say $X_\foo$ when \foo is a <mathchardef
token>.  So one of the requirements for upward compatibility with
plain TeX is that any control sequence that is currently a
<mathchardef token> must still be a <mathchardef token> in any
upgrade.

Is there a similar definition for AmSTeX?  Is it kosher AmSTeX to give
the input `$X_\nrightarrow$'?  Should we be prepared to let the odd
document containing `$X_\nrightarrow$' and the like influence the
arrow-building kit?  How upwardly compatible should we aim to be?

These questions and more will not be answered on this week's
installment of... Soap.

Alan.