[Fontinst] A classic: SC & OSF sets

[Philipp Lehman]

I didn't install any new fonts for quite some time and it seems like 
my command of fontinst got a little rusty. While I'm trying to write 
a fontinst tutorial, thinks that used to be clear are not so clear 
any more when taking a closer look.

Here's a 'classic' problem: suppose I'm writing a fontinst driver for 
Adobe Sabon, basic plus SC & OSF package. One peculiarity of Adobe's 
SC & OSF sets is that they omit upright OSF fonts for the regular 
weight and put the OSFs in the SC font. This requires exchanging/ 
renaming glyphs in three cases:

1) Creating an upright VF with OSFs (e.g. T1/psbj/m/n)
2) Creating a SC VF with lining figures (e.g. T1/psb/m/sc)
3) Creating an upright, TS1 VFs featuring OSFs (e.g. TS1/psb/m/n)

1) When looking at some old drivers of mine and searching the archives 
of this list, it seems to be common to do that as follows:

\installfont{psbrj8t}{psbr8r,unsetnum,kernoff,psbrc8r,kernon,latin}{t1}%
  {T1}{psbj}{m}{n}{}

What is puzzling me is toggling \setkern with kernon.etx and 
kernoff.etx. A couple of questions:

* \setkern does not overwrite existing kerning data, right? Do we need 
to disable \setkern at all in this case? After all, both psbr8r and 
psbrc8r feature the same glyph set.

* If we do, what about the kerning of the OSFs in psbrc8r? In this 
case, their kerning data is lost, isn't it?. What's more, 
unsetnum.mtx doesn't seem to remove the kerning data of the lining 
figures; it just clears the glyphs. So wouldn't we get OSFs kerned 
like lining figures here?

2) That might be done like this:

\installfont{psbrc8t}{psbrc8r,unsetnum,kernoff,psbr8r,kernon,latin}{t1}%
  {T1}{psb}{m}{sc}{}

The same questions apply...

3) is an interesting case as well:

\installfont{psbr8c}{psbr8r,unsetnum,kernoff,psbrc8r,kernon,resetosf,textcomp}{ts1}%
  {TS1}{psb}{m}{n}{}

* Same questions as above...
* What about resetosf.mtx with respect to kernon.mtx? Should \setkern 
be reactivated before or after resetosf.mtx is read?

Your comments are appreciated...

-- 
Philipp Lehman  <lehman@gmx.net>