[Fontinst] Bug in fontinstversion{1.927}

[Walter Schmidt]

On Sun, 2 Jan 2005 22:05:31 +0100, Peter Dyballa wrote:

> The 8r encoded MTX files do not contain all the OT1 or T1 or TS1 
>glyphs 

Indeed!  The 8r character set comprises a number of glyphs, which
are found in "usual" fonts from the non-TeX world AND which are 
useful to build the T1 or TS1 character sets.  (Btw, forget about 
OT1.)

>so it's close to nonsense to create an OT1 or T1 or TS1 based 
>MTX from such a set.

This conclusion holds only, if your "physical" font comprises 
already many glyphs which are useful for T1 or TS1, but are not 
captured by reencoding to 8r.
In this case you should reencode it _directly_ to T1 and TS1.
I.e., you must create two different (T1, TS1) reencoded instances 
of the base font, rather than only one (8r).  If there remain still
some missing glyphs which can be built from available pieces, you 
can build virtual fonts upon the two reencoded ones.  (NB:  These 
"final" virtual fonts must be given other names than the reencoded 
ones, even though they have the same encodings!)

The problem lies in the fact, that T1+TS1 together comprise more 
than 256 different glyphs.  If the physical base font has already 
more than 256 "useful" glyphs, the final T1 and TS1 fonts cannot 
be based upon the same reencoded font (e.g., 8r), because an 
encoding vector cannot have more than 256 elements.  

Walter