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

Re: math fonts, etc

Well, I think I could come up with my own set of three different lines
that the discussion here follows or should or shouldn't follow.

1. Implementation of the proposal of Justin Ziegler
2. Design of standard TeX/LaTeX control sequences to access all those
3. Discussion about missing glyphs, fonts, typographical traditions and
   rules etc.

I admit that I'm rather concerned with the latter two, but I fully
acknowledge Frank's point here that implementation of the proposal
should be the first step. Also, taking out some glyphs (at least at
first) seems to be a good point, as some glyphs are non-existent yet,
some are questionable as members of the core etc. Hopefully, this will
also help in having an implementation soon.

Within the confinements of TeX it seems hardly ever possible to have
an implementation meeting all needs. The ideal user would put together
some virtual fonts for the glyphs he needs with the proper kerning
between them, but this is very unrealistic, so I think the proposal
serves its purpose well enough.

I've seen too many books using Computer Modern only for math with another
typeface for text (strange enough, many of them typeset in Times), so
there is really a need to have math fonts for many different typefaces,
and this is the main goal of the proposal.

2. seems to be important and could go well along with the implementation,
and I think this should be discussed further on this list.
Many of the new glyphs could be accessed in a yet existent standard way
(e.g. font switches exist, only lowercase letters to be accessed by that
are missing in existing fonts; wider accents are only adding in
functionality), but there are many other new glyphs for which there isn't
yet a control sequence. Further, for using the upright/slanted rule
properly, there has to be a way to address the most frequent constants
falling under this rule (e, i, pi, d etc.), as these are rather frequent,
I feel that they should bet rather short control sequences. Without
a standard here, much of the implementation seems quite unusuable.
And discussing the implementation then when playing with it
will be hampered, as one has to give slot numbers or to explain one's own
set of macros to be understood by others.

  This standard could even consider control sequences for concepts
for which there isn't yet a special symbol (e.g. using \Laplace for the
Laplace operator and \Diff for a (forward) difference operator when both
are typeset as \Delta); this could be of interest for forward compatibility,
e.g. for textbooks where future editions could easily be set then with
a wider set of glyphs (in the second volume of a textbook that I'm currently
setting I'm using control sequences for intervals, but the output is
the usual (a,b), [a,b] etc. notation and not with special delimiters,
as the author wishes to keep that for compatibility with the first volume
(which hasn't been set by me); a future edition could easily change this).

  Without a standard, any future manual concerned with math typography
in TeX must tell the reader something like "The usage of these rules and
symbols is strongly recommended. But TeX doesn't offer a way to
use them to give you the possibility to design appropriate macros for
yourself". Most users would be very glad for this opportunity to become
an experienced macro programmer and to spend some days or weeks to get
the needed set of macro, rather than to write the paper they intended.
[I'm sorry, but partly this is intended especially for Hans Aberg, who
opposes the proposition of standards on the one hand but complains about
TeX's difficult usage on the other]

  An experienced user will of course always introduce some of his own
macro names, e.g. using \let\tensor=\otimes when using this symbol for
notation of tensor product, but it is a great difference to use essentially
a set of \let commands (with a standard) or to create all the macros from
scratch, involving searching the correct font and slot number, choosing
the right math symbol type etc (without a standard).

3. clearly doesn't belong here, but to the math font discussion list.
To collect all the information in these fields will certainly do no harm
to LaTeX, but neither it does much service at the moment. We should have
a compatible standard first; adding more encodings, more fonts and glyphs
could come at a later stage. At present, we could simply collect all
propositions, all glyphs, metafont programs etc. May be glyphs yet done
could be put together in a mere "glyph container" (to use a term from Omega),
so other users could see them and comment on them, without having a proper
enconding let alone kerning information etc at first. Later, this could
result in having some extra LaTeX-cm-mf-only non-standard fonts,
so that most users could use LaTeX with the proposed implementation,
but with the font family they like (hopefully); a mathematician demanding
special symbols or going for an extra bit of typography would have
to use Computer Modern with the extra fonts then.

Johannes Kuester

Johannes Kuester                    kuester@mathematik.tu-muenchen.de
Mathematisches Institut der
Technischen Universitaet Muenchen