[pdftex] TeX as a composition server?

Robin Fairbairns Robin.Fairbairns at cl.cam.ac.uk
Sun Oct 24 00:04:34 CEST 2010

ivo welch <ivowel at gmail.com> wrote:

> I would suggest that you try out how fast tex can produce your sample
> documents.  it is not hard, and more informative than guessing.


> from what I can see, as an ordinary end user, TeX can do almost any
> font, but it is not seamless.    the luatex vs. xetex vs pdftex
> specializations are painful.  really.  every tex should come with only
> one version that preloads all system fonts, all latex fonts, and make
> it easy to load fonts from a different location.
>   \addfontdir{c:\windows-yikes\morefonts\}
>   \loadfont{crazyfont}
> preferably, tex should work out for itself standard encodings,
> metrics, etc.  of course, the fact that it is not so is not really
> latex's fault, but the fault of the font mess in the real world.

the problem is that until (relatively) recently, tex has stood still
w.r.t. font technologies.  it was (pretty) easy to map type 1 format,
once it was published, but there wasn't any really usable attempt with
ttf until xetex came along.

> I would love to find a way to produce epub files from latex or context
> files.  yes, typesetting each word as an image does not sound like a
> great idea.

no-one has ever managed a "foolproof" way of generating any sgml-based
format from latex, afaik.  the likeliest candidate would seem to be
tex4ht, which uses tex to do its parsing, but even that doesn't do the
whole job (and probably doesn't have an epub backend -- i've not

> besides, it would be good if TUG assumed a more proactive role, and
> officially deprecated old aspects of TeX.  time to move into the 21st
> century.  time for luatex to become the official standard at some
> point.  I know it may break compatibility, but TeX's age is really
> beginning to show.  the number of hackers that can still program in
> this ancient tongue is not growing.

what would be gained by that (apart from antagonising huge numbers of
the existing user base)?

tug, in essence, has no _power_; if tug is going to achieve anything, it
needs to be achieved by exhortation of the user base to go out and do
that thing.

and while luatex may be pretty good just now, iirc the luatex team don't
expect it "finished" until 2012.  dropping tried and tested technologies
in favour of luatex just now would (imo) be counter-productive, quite
apart from the alienation mentioned above.


More information about the pdftex mailing list