I agree mostly with Sebastian (and the others) that XSL-FO is the (probable) future, but on a pratical level, for documents that have a lot of scientific content, mainly math, if we want to print something typographically acceptable, going the (La)TeX route is still (for years to come, awaiting the inclusion of math typesetting into XSL-FO) the only real possibility. I am actively working on---and I am particularly interested in help with---finding the best way to marry DocBook (+Math, SVG) with print, be it via XSL-FO (a lot of tuning remains, think about table handling, not mentioning math), xmltex (directly making TeX interpret the XML, probably a non-starter for complex DTDs such as DocBook), or the XSLT to LaTeX approach of db2latex. I think we should work on all of them in parallel, so that we can learn the difficulties along the road, and pick the solution that works best today for the problem at hand, improving at the same time the probability that a global XSL-FO-based approach will see the light one day.

Urgent work to be done today is:

1) Complete the XSL-FO part of Norm's stylesheets (as I said, tables need some work, since TeX's table model imposes constraints that make a general implementation quite involved). 2) expand xmltex (especialy the math bit) 3) complete db2latex (also in particular the math bit)

I think that a coordination of the effort spent in all of these areas would be worthwhile to optimize resources.

All the best, m