MIME-version: 1.0
Content-type: TEXT/PLAIN; CHARSET=US-ASCII
Content-transfer-encoding: 7BIT
Mail-System-Version: <MultiNet-MM(369)+TOPSLIB(158)+PMDF(4.3)@MATH.AMS.ORG>
Message-ID:  <832684833.315163.MJD@MATH.AMS.ORG>
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@relay.urz.uni-heidelberg.de>
In-Reply-To:  <199605192340.JAA09537@axiom.maths.uq.oz.au>
Date:         Tue, 21 May 1996 09:20:33 -0400
From: Michael Downes <MJD@MATH.AMS.ORG>
Sender: Mailing list for the LaTeX3 project
              <LATEX-L@relay.urz.uni-heidelberg.de>
To: Multiple recipients of list LATEX-L
              <LATEX-L@relay.urz.uni-heidelberg.de>
Subject:      Re: More on fontdimensions
Status: R

> J"org Knappen <KNAPPEN@VKPMZD.KPH.UNI-MAINZ.DE> writes:
>
> > I have now a test implementation with the additional fontdimensions above
> > Nr. 22. Unfortunately, this causes the tfm files and the space needed in
> > the font memory to grow more then necessary, because all empty
> > fontdimensions are automatically set to zeroes.
> >
> > Therefore I'd like to ask, if this waste of space is tolerable, or whether
>                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

[Ken Smith:]
> Surely today we don't need to worry about wasted space on hard disks,
> and the speed of modern machines is such that scanning a tfm file for
> the relevant dimension(s) adds a negligible amount of time in processing.

Yes, that's true, but the question that should be answered first is
whether there is *any reason at all* for wasting the space. I don't
think there is. The only motivation that I can recall from earlier
discussions was some idea of avoiding overloading of the fontdimens
8--22. But that doesn't sound very reasonable to me, since 8--13 are
already overloaded anyway: cmsy and cmex type fonts use them for
different purposes.

So in order to show that an additional fontdimen should go in position
23 or above, it seems necessary to argue that the fontdimen might be
useful for a math symbol font such as cmsy. I would say none of the ones
in J"org's list qualify

> 23: font_cap_height      % Height of non-accented caps
> 24: font_ascender_height % Height of lowercase letter with ascenders
> 25: font_accented_cap_height % (Maximal) Height of accented caps
> 26: font_descender_depth     % (Maximal) Depth of lowercase letters
> 27: font_max_height   % Height of the highest object in the font
> 28: font_max_depth    % Depth of the deepest object in the font
> 29: font_digit_width  % Width of the widest digit
> 30: font_cap_stem     % Breadth of the stem of capital I

because they make sense only for text fonts. All of the symbols in the
cmsy encoding are designed for use as individual objects in a math
formula; therefore measurements across a group of symbols for cap
height, max depth, etc ... well, they may be interesting but they are
never going to be needed for use in TeX macros!

(Furthermore, if you really want cap height, etc for the calligraphic
alphabet that is embedded in cmsy, the right approach IMHO is to move it
to a different font, not to go further down the road of making fonts
such as cmsy serve simultaneously as symbol fonts and text fonts. Is not
the embedding of the calligraphic letters in cmsy plainly artificial?
The font handling constraints that Knuth was working with back in 1980
or so induced him to pack as many math symbols as possible into four
fonts of 128 chars each. But the alphabet clearly has no design traits
in common with the rest of the font. The same could be said of the
blackboard bold alphabet in msbm.)

However:

Although font_cap_stem doesn't make sense, as defined, for a math symbol
font, it might be worth having under its real intent as a measure of
`absolute boldness'.

You could also argue that font_max_height and font_max_depth might prove
useful somehow for a math symbol font. But I would like to see someone
give an example of a real application first.

> I recently installed emTeX on a Pentium for students in this Department.
> The machine has a 1.2 Gb disk, and each of the tfm files took up 32 kb
> since DOS doesn't allow more than $2^{16}$ files in any partition, and
> the disk was configured as a single partition.  With files taking up
> this much space I don't think you need worry about some extra dimensions
> in some fonts "wasting" space.

The whole purpose of emTeX's .fli files is to save the accumulated
wasted space at the end of large numbers of small files.

Michael Downes
mjd@ams.org