Hello All,

Vladimir writes:

 > In the file test-utf8.tex, i use commands like
 >
 > \fontencoding{T2A}\selectfont
 > \fontencoding{T1}\selectfont
 >
 > but you can add definitions like
 > \def\setcyr{\fontencoding{T2A}\selectfont}
 > \def\setlat{\fontencoding{T1}\selectfont}

That leaves us switching between two environments
rather than the expected three.

Having done this sort of programming, I would
guess that all switching could be eliminated via a
more complex UTF8 macro package.  But it would be
a feat of macro engineering that I am unlikely to
undertake.  (Perhaps others too so long as the feat
is considered damaging to good typography.)

In these circumstances, I believe that Ruprecht
should also try Omega (the TeX extension). It is
fairly widely available (see CTAN?).  In it, I
believe a single 16bit virtual font using unicode
contains all the necessary characters. Quite
frankly, Omega is probably the one full solution
available. The software engineering involved here
was to hoist TeX sourcecode into the world of 16bit
fonts.

It would be good to hear concrete first-hand
performance reports!

Cheers

Laurent S.

PS.  Finally, one can build virtual fonts with a
private 8-bit encoding for Russian/Polish/German
(ONLY!); 256 is plenty of characters.  The results
would be optimal but hard to generalize.  (And you
would win no prizes;-).