Hans Aberg wrote:
> In fact, I am not sure that even the things that Frank Mittelbach wants
> to do with LaTeX3, can be fully done without requesting a new TeX version.
> This importance of the LaTeX3, namely, as an indicator on how TeX might
> evolve, was already recognized by the NTS project several years ago.
This is precisely what I have observed in many (or all?)
discussions on this list. So I'd like to make an
unqualified comment (being aware that this has probably been
discussed internally by the LaTeX3 project members, and
that Frank Mittelback prefers to talk only about
Knuthian TeX in this place):
All the more technical issues discussed seem to eventually
boil down to some basic limitation in TeX the program and
a lot of time and effort is spent by mending these on the
macro level, or by reencoding fonts.
The most striking example might be the non-hyphenation of
words containing accented letters. To get around this problem
which might be have been fixed by extending TeX, the
EC fonts (thanks Joerg!) were developed in a probably
immense effort. Although they undoubtely improve CMs
appearence with non-English texts, they also perpetuate the
division of the world into TeX and non-TeX. Following
some discussion in comp.text.tex, it seems to me that
it is unlikely that postscript versions of the EC fonts
will appear any time soon---bad for web based TeX
publishing, and for every commercially available
font someone will have to at least T1-reencode it,
which I guess is not a triviality. So TeX users will
always lag behind in the choice of fonts they can
use (if they want to avoid running into the non-
From the user perspective (I really don't know anything
about the internals and the philosophies behind them),
it simply amazes me that the main trust of the LaTeX3
project is not directed towards fixing such an
elementary and apparent flaw within TeX. With all due
respect for Don Knuth, TeX appears to be designed in a
very anglo-centred way, and I don't see why hypothetical
version of LaTeX should perpetuate this.
Much of the mathfont problems seem to be due to some other
limitations in the font handling in TeX, I also remember
some discussion about shortref mechanisms, which
essentially got to the point "it cannot really be
done in TeX, but we can write some clever hacks to do it".
Unfortunately many packages contain already too many
clever hacks (just thinking of amsmath: extremely useful,
but also full of minor bugs and inconveniences that
seem to be difficult to fix).
Is this just my impression, or do the experts see