# Re: defining the script sizes via font dimens

• To: math-font-discuss@cogs.susx.ac.uk
• Subject: Re: defining the script sizes via font dimens
• From: Frank Mittelbach <Frank.Mittelbach@Uni-Mainz.DE>
• Date: Wed, 22 Oct 1997 23:20:07 +0100

Matthias Clasen writes:
> > that interface would have the advantage that it would be pramble only
> > thus could be as complicated and space consuming as you like.
>
> Well, I have not yet done any experiments with that idea, but wouldn't
> this have the side effect of loading \textfont1 in all sizes for which
> such an \DeclareDefaultMathSizes statement is given, even if we never
> set a formula in that size ?

true but this would not be too much of a problem as it would only be
one font not the whole set of math families.

> for stuff that does not scale linearly is causing more problems than it
> solves. (And I wonder if this is not already a problem for mathptm and
> exscale, since many of the extra font dimensions from appendix G probably
> do not scale linearly as well)

with exscale i have my doubts as well and i think i mentioned
something like this in its documentation

> I tend to revert to the standard LaTeX interface for script sizes and
> math spacing. It will be easy enough to provide this information in
> newmath.sty for every layout which needs different values here (currently
> this is only mathptm).

all in all perhaps a wise step :-) which i somehow voiced when i
talked about "worth the trouble". after all there is one conceptional
problem: while those values are somehow related to font families or
say set of fonts they are not really depending on a single font. true,
that's the way Don implemented all his parameters for math, but then
one has to recall that originally he was implementing just one set for
one size and this generalization wasn't in his mind. so it seems to me
a bit questional whether or not we should tie them to each
font. therefore tying them to the set of font (via an initialization
file, for example) seems to me logically the better approach even if
that has disadvantages of its own (like the need for such an extra
file).

> Another approach might be to invent a new
> filetype (perhaps mfd - math font description. This would at the same time
> give proper credit to this mailing list :-) like fd-files containing this
> information, one file per math layout. We might thus have default.mfd,
> mathptm.mfd, etc. Is this a sensible idea ? Perhaps even that would be
> overkill.

i don't think this is overkill. the only disadvantage i see is that it
means one extra file extension. and if it turns out that in 99 percent
of the cases nobody ever changes anything you might as well put the
defaults into one package file. but it would be the right level of
abstraction in my understanding

frank