TeX extended fonts

[J%org Knappen <KNAPPEN@VKPMZD.KPH.Uni-Mainz.de>]

This post contains in its forst part some answers to Peter Dalgaard, in the 
second some more general things.

First of all, the Cork standard is not only a pure prescription, there 
already exist some implemantations of it:
--The dc-fonts by N. Schwarz, Bochum
  They are written in METAFONT. 
--The em-fonts sold by ArborText. They are virtual fonts. Some characters 
  (Edh, thorn, eng) are missing.

There are also dc-hyphenation patterns available for several languages, but I 
don't know if danish is among them yet.

Somewhere I have a TeXed table which simulates the Cork font layout using 
cm-fonts. I can send it, if you want. The table is worth careful study 
before demanding for changes or other font encodings.

                      Now to the general part:

One of the main advantages of TeX is its high degree of portability. This 
would be immediately destroyed, if we had fonts with free slots, which are 
different from country to country. Noone will have all the national fonts 
ready at their installation.

The Cork (ec) scheme is a good compromise and I think it serves its purpose 
(typesetting text in the most european languages\footnote{american english 
is considered a european language}) very well. There are some 
shortcommings, of course.

  Non european languages have additional letters (vietnamese, african 
  languages, amerindian languages) to the latin alphabet, which are not 
  covers in the ec scheme.

  Special character for old english (yogh, wynn, long s, long s ligatures)
  are missing.

  Some european languages (lattvian, celtic languages, maltese, sami) are not 
  covered. 
  
  The formation of ligatures is language dependent, e.g. you don't want an
  fi-ligature in turkish.

A solution of these problems might be the following\footnote{I got this idea 
>From Yannis Haralambous}:

Create a font, gc (generic computer), which contains all the elements to 
build the characters for latin writing from. These elements contain base
letters, ligatures and diacritics. Build virtual fonts from these elements. 
Maybe we need two of them, because there are more than 256 ``elements'' to 
consider. 

IMHO, the gc fonts should be a supplement to the ec fonts and contain only 
elements not included there. I wish it could be possible to keep the number 
of virtual fonts and encoding schemes low, but I'm afraid this is an 
illusion. Everybody likes to emulate their favourite machine character 
sets :-(. For portabibility it is necessary to resolve the virtual fonts 
into its real components (using an utility like P. Breitenlohner's dvicopy).

Yours sincerly, J"org Knappen.

P.S. Karl Berry's message ended with the line
``Here is the complement font:''. I'm afraid, that it was mangled. Karl,
could you please resend the rest of your message?

P.P.S. (to Tom Ridgeway) In the fc (aFrican Computer) fonts I used a similar 
ansatz. By rearranging the program files it is possible to save cpu time
(the base letter is computed one time and saved as a picture, only the 
diacritics are added). The fc fonts are available from 
ftp.rus.uni-stuttgart.de  under soft/tex/fonts/metafont/fc