Re: Font naming rears its ugly head again

[mackay@cs.washington.edu (Pierre MacKay)]

In virtually all the text fonts that are available, with the very specific
exceptions you have mentioned, it is clear from a quick inspection of the
PFA file that the composites mentioned in the AFM as sent out by
the foundry are composites in the PFA file as well.  Unless the
foundry is trying to be deliberately misleading, when it specifies
in its own AFM file

CC aacute 2 ; PCC a 0 0 ; PCC acute 42 -8 ;

it is essentially putting in a human-readable form the
very same information that is coded into unreadable
binaries in the PFA file.  There are only 16--18 bytes of binary in
any of the composite recipes I have seen; four of them are
invariant for the entire class of composites, and four more for
any group of accented letters.  It doesn't leave much remaining for
the effort of repositioning accents and recording the width.

So I am lead to the conviction that what is specified in the PFA file
is essentially the same as what is specified in the VF file.
If it isn't, the conclusion must be that the information in
the AFM file is in some sense a lie.  I hope I doubt that.

I don't understand entirely your strictures about dvips.  
dvips does what it is told to by the VF file, and the
VF file contains a translation of what was in the AFM.  It may be that
at some point there have to be grid-fitting adjustments,
but since those are presumably taken care of by the hints
in the simplex characters I really doubt that there is
any change between the way dvips lays down a composite
and the way the postscript interpreter lays down the
same character from the PFA file.  I haven't had
time to make the experiment, but I might try setting
some composites at 96 points with the VF file and the same
characters directly from the PFA.  300dpi is coarse
enough that one-pixel offsets might just be visible,
but I shall be astonished if there is anything worse.  

Allow me to stipulate for the argument that the composites
generated by the VF mechanism are the same as the composites
generated by the recipes in the PFA.  If that is the case
I can save a considerable amount of space in display bitmaps
for xdvi or dvipage by leaving out the six redundant bitmaps
for the letter a.  Given the size of available font
resources even in the public domain fonts, the saving is significant.
For the PC users I am trying to prepare files for, it mattters
even more.  That is a constraint that is not likely to go
away in the near future.


Email concerned with UnixTeX distribution software should be sent primarily
to:	elisabet@u.washington.edu	        Elizabeth Tachikawa
otherwise to:  mackay@cs.washington.edu		Pierre A. MacKay
Smail:  Northwest Computing Support Center	Resident Druid for
	Thomson Hall, Mail Stop DR-10		Unix-flavored TeX
	University of Washington
	Seattle, WA 98195
	(206) 543-6259