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

Re: About atomic encoding



>Using an atomic encoding for real fonts would thus allow producing
>customized versions of the core font from the same METAFONT source

Remember that the majority of published technical documents don't use,
and probably never will use, MF fonts.  The AMS is a notable
exception, but even they use Autologic Times for the math core font
(I'm sure barbara will correct me if I'm out of line!)

>In any case, it would be necessary to produce different sets of 
>TeX macros to access these characters, but most of them would 
>be the same in all variants and they could be produced from one 
>single source file using different DOCSTRIP modules.

The problem with this approach is that the macros need to change
depending on the math version.  For example, a chemist might type:

   \documentclass{article}
   \usepackage{chemistry} % NB: This is a fictional package!
   ...
   foo bar \( 2n + \sum_i x_i \) foo bar
   foo bar \< H_2 + O \rightarrow H_2 O \> foo bar 

where \( ... \) selects math version `normal' and \< ... \> selects
math version `chemistry'.  If the math core fonts in the different
math versions use different encodings, then either swapping between
the math versions will require redefining a lot of macros, or some
macros will have to depend on the current math version.  Either way,
there will be a heavy time penalty.

Alternatively, the special symbols for chemistry could be put in a
separate font, loaded by the chemistry package.

So there's a trade-off between speed and font usage.  Experience with
non-standard text font encodings (such as those produced by afm2tfm)
has demonstrated that allowing variant encodings makes the
package-writer's job much much harder.  (One of the original reasons
why I developed fontinst was because I was very very tired of trying
to maintain font packages where different fonts had different
encodings.) 

>from a physicists point of view: I'm not completely happy with
>the idea of having (a) a math core font with italic letters 
>and two sets of greek (italics and upright) and (b) a math 
>operator font with upright roman letters and no upright greek. 

I'd be unhappy too if that was the suggestion!  But it's not.  There
is *no* specification as to the shape of the math core or the math
operator font (OK, apart from the Greek letters, which are still under
discussion).  So a chemist could use a math core font containing
upright glyphs, and a theoretical computer scientist could use a math
operator font with upright sans.

So in your example, the `particle physics' math version can contain a
math core font containing upright letters.  If you want math italic in
the same formulae, fine... there's no restriction saying that a math
version can have only one MC-encoded font.  The only restriction is
that you have to pick one of them to be \fam2, in order to set the
fontdimens appropriately.

I agree with your analysis of the problem, it's very important to
allow different diciplines freedom about the shape of the math core.
I just think that allowing different encodings isn't the way to solve
things! 

>Any opinions? Perhaps this might initiate some discussion about
>math fonts here again...

I think it might :-)

Alan.

Alan Jeffrey         Tel: +44 273 606755 x 3238         alanje@cogs.susx.ac.uk
School of Cognitive and Computing Sciences, Sussex Univ., Brighton BN1 9QH, UK