[XeTeX] The future of XeTeX

Zdenek Wagner zdenek.wagner at gmail.com
Wed Aug 1 10:31:10 CEST 2012

2012/8/1 Simon Cozens <simon at simon-cozens.org>:
> On 31/07/2012 18:06, Keith J. Schultz wrote:
>> Lua(La)TeX is a move in this direction. Modernizing TeX!!
> Well, yes and no.
> The problem with all the TeX engines, the elephant in the room that nobody's
> talking about, is TeX itself. Don't get me wrong, I think it's a great piece
> of code, and it's done fantastic service over the past thirty years, I use
> it every day, and I don't see myself using anything else in the near future,
> but...
> Let's just say there's a good reason why there isn't a great deal of code
> that's still in use after thirty years.
> When Don Knuth wrote TeX, he needed an embeddable programming language, a
> font specification system, a font library, an output format. Absolutely none
> of those things existed at the time, so he wrote all of them from scratch:
> the TeX language, MetaFont, TFM, DVI. Given that he was pioneering and had
> nothing else to draw upon, he did pretty damned well, but over time they
> have not exactly turned out to be the best choices.
> *TeX development since 1982 has essentially been a bunch of disparate
> projects, each trying to rip out something that Don did and replace it with
> something more sensible instead. So we had NFSS and virtual fonts to remedy
> the deficiencies of MetaFont; then we had pstex and pdftex to remedy the
> deficiencies of DVI; xetex to remedy the deficiencies of TFM; luatex to
> remedy the deficiencies of the TeX language.
NFSS is not a hack. If you see it like this then everything beyond the
TeX primitives would be a hack. NFSS is just a macro package that was
in nineties incorporated into the LaTeX kernel.

Virtual fonts do not remedy METAFONT deficiencies. In fact, if you
build virtual fonts from MF sources, you can achieve exactly the same
result with pure MF. Virtual fonts have to remedy the deficiencies of
encoding. The Latin alphabet for all languages (including Vietnamese)
contains 500+ characters but the non-Unicode encodings allow 256.

XeTeX is not a remedy to TFM deficiencies, it is again the remedy of
encoding deficiencies. It won't be that difficult to extend TFM but
implementation of Unicode and support for comlpex Arabic and Asian
scripts required complex rewrite.

> They've all been great hacks, but they've all been hacks.
> My feeling is that it's time to accept the principle of "one to throw away"
> and finally put TeX82 out to pasture. Now we are blessed with a set of
> technologies which have proved themselves, which give great results on
> modern systems and have support for problems which were not even on the
> agenda thirty years ago. Just take your favourite scripting language, your
> favourite shaping engine, and your favourite output engine, stick the
> Knuth-Plass box-and-glue model, justification engine and page builder in the
> middle,  glue them all together, and call it something new.
> Simon
> --------------------------------------------------
> Subscriptions, Archive, and List information, etc.:
>  http://tug.org/mailman/listinfo/xetex

Zdeněk Wagner

More information about the XeTeX mailing list