Re: psnfss and lw35nfss

["Peter Harmand" <harmand@math.uni-oldenburg.de>]

>  >  >  b) are the kermings in 8r.mtx bad, as Thierry suggests
They are very bad for German typesetting: the kerning pairs T-udieresis 
and W-odieresis are quite frequent (eg. in doors and words). With fonts
which use a lot of kerning, like Bitstream's, the result looks terrible.
 
>  > 
>  > I reckon so.  The kernings in latin.mtx are the best list I could
>  > manage...
> why has no-one said this for 2 years? i see it has Constantin's name
> on....
Because somewhere it is said that 8r is TOTALLY UNSUPPORTED and nobody
is using it. And the stake is high (at least for a beginner):
  * write 8renc.def (and notice that you lost all support by Them)
  * realize the kerning problems in 8r.mtx (no \latinfamily for 
    fontinst, do everything by hand)
  * realize that fontinst isn't well prepared to fake small caps in 8r
    (write 8rc.etx)
  * realize the following warning if you finally try
      \usepackage{times}
      \usepackage[8r]{fontenc}    
    ..  
    LaTeX Font Warning: Font shape `8r/cmr/m/n' undefined
    (Font)              using `8r/ptm/m/n' instead on input line 87.
    ..
I gave up at this point -- even though this is merely an unwanted warning. 
(But I'm still interested in a solution. The command \rmfamily\selectfont 
between the two \usepackage lines avoids the warning, but I wanted to use 
8r like T1. A \rmfamily before \fontencoding\encodingdefault\selectfont 
in fontenc.sty would also do. But I don't know if this is a sensible 
change request.)

However, more important for giving up was that as a texadmin you have to 
provide OT1 and T1. A ligful 8r doesn't save tons of tfm's and vf's. 
(The only thing I might miss are the original \ldots.)

When I made my remark about the missing ligatures in the 8r-encoded
metrics in the psnfss-beta distribution, it was really meant as a
question: Should one abandon the idea of a ligful 8r-encoding?
   
Peter