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

Re: Familiy limit (was: About atomic encoding)

Alan wrote:
> >This would however require loading another
> >math symbol font, which would be a waste of families
> Insert standard grumble about DEK's 16-family limit here...
> If you're keen to get round the family limit, you could try producing
> a MC-encoded roman font (suitable for use as particle symbols,
> checmical formulae, etc.) but with enough kerning information to make
> it possible to use it as an operator font as well.  It may be that
> this is technically infesible (the kern table might end up being huge)
> but you never know...

Concerning the family limit: I estimate that as a physicist I would
end up with around nine families: the six standard families and three
extra copies of the core font for math alphabets. The standard core
font would be math italics, the others would be math roman (for upright
core letters and particle symbols), bold math italics (for vectors)
and bold sans serif italics (for tensors). American publishers might 
choose to replace the last two with upright bold math roman and bold 
sans serif, but they wouldn't need much more families either.

Having nine families allocated would leave seven more free, thus there 
should be no reason to worry, unless several more families are wasted 
for typewriter or slanted fonts which are not normally needed in math. 
(Besides, is there any recommended way to undeclare math symbol fonts?
I could hack the allocation counter and \let symbol fonts to \undefined,
but is this OK or would it leave some internal NFSS tables cluttered up?)
Anyway, if we allocate only those families that are actually used in
math, the limit of 16 fonts isn't as bad as it seems at first.

Concerning the usage of math roman vs. text roman: I think the problem
is that a math roman (with math mode spacing) is only good for single-
letter symbols, such as upright core letters or particle symbols but
not necessarily for chemical symbols or physical units, which may also
consist of multiple letters. If the math roman font is to be used for
these purposes as well, you would need quite a lot of kerning pairs
to mimic text spacing in certain letter combinations. Fortunately,
there are only 109 or so chemical elements and only a limited number
of physical units, but I don't think that this is the optimal solution,
since it might yield e.g. `NaCl' with text spacing between `N' and `a'
and between `C' and `l' but math spacing between `Na' and `Cl'...
Physicists might like this idea because it would give additional 
visual clues about the structure of chemical formulae, but I doubt
if chemists really want it like this...

Any chemists here, who wants to comment on this?

Ulrik Vieth.

P.S. Alan: would it be possible to add a reply-to field to the mailing 
list address before any reply-to fields coming from the sender? I have 
been getting multiple copies of messages in reply to my postings and 
when I hit reply in elm, I get the sender address rather than the 
list-address by default, so I am sending multiple copies as well, 
if I simply add the list-address as cc.