[tex-live] free software, DFSG

Olaf Weber olaf at infovore.xs4all.nl
Fri Jun 4 20:12:40 CEST 2004

David Kastrup writes:

> If he is going to fix the remaining bugs of TeX (instead of declaring
> them "features"), it is going to take years.  Maybe they'll never get
> fixed at all.  Where do I turn if I need new features to make it into
> TeX?  The eTeX team?  Who is that?  Peter?  To Han The Thanh?
> Guiseppe?  John?  Karel?  Michael?  The main "advances" nowadays
> happen outside of TeX proper, as in kpathsea.  Which deals with the
> problem of most conveniently collapsing an artfully structured
> directory tree into a flat namespace.


> No, a Knuth cult is not the problem.  What is hindering the TeX world
> from progressing is a TeX cult, and a good-enough cult.  We need to
> leave TeX behind.  And Knuth is doing his part in that: he has nailed
> TeX down and has stated repeatedly, definitely, that it is not going
> to move anymore.  It's just that nobody listens to him.

This week I spent some time investigating the possibility of having
spaces in filenames.  The simple case is done, in that

	\input "foo bar"

works about as well as you'd expect.  The hard part is making sure
that (for example) LaTeX's

	\input{foo bar}

would also work.  It turns out to be surprisingly hard to get this
right even in most cases, and then you need to tack on another bunch
on heuristics so that things like

	cmr10 at 5pt

aren't interpreted as a single filename.  I decided that it wasn't
worth the trouble and risk of introducing obscure bugs, so

	\input{"foo bar"}

is what you'll have to use in LaTeX.  My conclusion is that if you
really want spaces in filenames in a natural manner, the current TeX
internals are the wrong place to start from.

I really admire Han The Thanh for doing pdfTeX with TeX as the
starting point.

And I personally believe that this can be generalized: the algorithms
TeX uses for typesetting are still very good and useful.  However, it
would be a mistake, IMAO, for a successor to TeX to be based on TeX's
internals.  I'd like any such hypothetical system to be fairly
modular: a central typesetting core, with different front-ends for
various input languages (TeX's macro language, XML, ...) and different
back-ends for various output languages (PDF, PS, DVI, XML, ...).  This
is why the actual NTS project was a bit of a disappointment to me.

> I am working with all of the junk.  Who do I ask for new stuff?  I
> want John to get Omega into useful state.  It is a waste of resources
> to have LaTeX, PDFTeX, eTeX and others all forked off from something
> which does not even deal with character sets in a sane way.
> Guiseppe's Aleph is at least an attempt to get something workable
> again, but of course it misses PDFTeX functionality.

> And, of course, all of the stuff is impossible to maintain.  Add a
> feature and recompile?  Good luck.

> Maybe one day TeXlive will come without TeX.

Olaf Weber

               (This space left blank for technical reasons.)

More information about the tex-live mailing list