Re: T1 encoding suddenly used

[Karl Berry]

Hi Norbert,

"Suddenly"?  Nothing about this has changed in texinfo.tex (or anywhere
else I can think of) for years.  But I suppose if you were using the
texinfo.tex from 4.13, ca.2008, (despite my continual pleas to everyone
to always use the current released texinfo.tex from ftp.gnu.org :), you
would "suddenly" see a change.

texinfo.tex does use the EC fonts to typeset a few stray characters --
including the non-typewriter braces, since 2010-04-18.  It may be that I
could take those from cmsy/cmbsy, which I guess would be preferable;
I'll have to go back and look if there were issues.  It's certainly more
trouble to take them from cm*sy.  (FYI, the other characters are things
like eth, thorn, the ogoneks, etc., which don't exist in CM at all.  You
can search for \ecfont to see a list.)

As for the T1 encoding, texinfo.tex knows nothing about font encodings
in the LaTeX sense.  The T1 fonts are being pulled in simply because
that is what psfonts.map/pdftex.map says to do for the basic EC fonts:
ecrm1000 SFRM1000 " T1Encoding ReEncodeFont " <cm-super-t1.enc <sfrm1000.pfb
For that matter, I don't see any non-T1-encoded entries for ecrm* in
psfonts.map.

It is not the case that "most texi files" use T1.  What is the case is
that if any texi file uses certain glyphs, those glyphs come from non-CM
fonts.  FYI, the "feynman" fonts are used for @euro{} (feymr10, etc.).
(And lcircle10 for @cartouche{}, but I doubt that one is an issue.)

Hope this helps,
karl

P.S. The general problem of encodings (both input and font), non-Latin
support, etc., etc., in texinfo.tex remains unsolved.  My friend Oleg
was getting close to a workable font system, but then stopped working on
it.  It's in the contrib/texifont/ subdirectory of texinfo if any TeX
hackers get interested.