Re: Font metrics in AFPDS?

[John Yeung <gallium.arsenide@xxxxxxxxx>]

James, I am impressed by your typesetting knowledge.  It's clear
you've done at least as much reading as I have, if not done
typesetting work yourself (which I haven't).

On Wed, Mar 18, 2015 at 8:56 PM, James H. H. Lampert
<jamesl@xxxxxxxxxxxxxxxxx> wrote:
> I actually have manually generated width tables (for Postscript fonts that I
> wanted to use in Xerox Ventura Publisher, but for which there weren't any
> compatible width tables), and well, there are only 190 characters to deal
> with in EBCDIC, and around 50 of them are just diacritical variations on
> others, so it wouldn't be all that difficult to do it myself. But if
> somebody else has already invented that wheel (and at some level, somebody
> HAS to have, in order for the fonts to be usable), it's a shame to have to
> reinvent it myself.

I hear you.  I understand what you are talking about.  In my case, I
manually generated width tables for use in calculating Excel column
widths.  (Well, semi-manually, since I could leverage some
programming.  I like to call it *empirical* generation of width tables
in the sense that I got the values by experimentation.)

And you are completely right that, obviously, at *some* level, someone
has done this work.  But one of the problems we face is that different
implementations of the fonts have different widths.  So the same-named
font rendered on two different printers or screens can be different.
Then there is the substitution problem, where (in some contexts) a
requested font may not be available, and the rendering engine
substitutes an available one instead.  And then of course there is
kerning and such if you need very high precision.  (That last one
means that the tables are not as simple as most people think; it's not
like each character has one number that expresses its width and that's
that.)

I'm sure none of that has escaped your consideration, so chalk it up
to me putting it out there for the archives.

The bottom line is that you're unlikely to find the information you're
looking for in already-published form.  You most likely will have to
build it yourself empirically.

John Y.