Whither TeX in the XML Régime?

(WARNING: The following prose
employs rhetorical flourishes
alien to a strict Strunk-and-White aesthetic.
Reader discretion is advised.)

In the beginning was the manuscript, the refined product of two unlike minds - its author a creator, a firebrand; its editor cool, needle-sharp, a craftsperson, a diplomat.

In the end, there was the printed page, the unobtrusive, even transparent receptacle of its author's meaning - a page produced for the machine-mad, computer-crazed marketplace, and yet reflecting, in its well-placed folio, headline, and margins, in the fine kerning, ligatures, and leading of its print, the accumulated wisdom of a five-century typographical tradition.

How, then, did the manuscript become the page?

The metamorphosis was worked with TeX, a software developed for scientific typesetting by Donald E. Knuth, Stanford professor in severe mathematical disciplines. And Prof. Knuth would have the world call the TeX worker a TeXnician - for in the spoken language "TeX" ends with the chi phoneme, not xi.

(a) When toiling in the mid-1990s (a gentle era, graced by the cool, dim dawn of Linux), our TeXnician began with a humble ASCII text manipulator, making of the raw manuscript file, foobar.txt, a markup-enhanced foobar.tex. But this was markup with a vengeance, orders of magnitude subtler than the markup provided in the plebian parlance of HTML. Over, above, and beyond the inevitable incantation for invoking and dismissing italics, there was the power to invoke and dismiss the slanted Roman - a style with which the italic is by no means to be confounded. There were codes for the thinnest of spaces, for the most exotic of diacritical marks, for the most exacting of partial differentials and multiple integrals in the author's mathematical displays. Indeed, the TeXnician rejoiced in the assurance that even were the all-but-unthinkable to occur, even were his firebrand author to abandon physical science for the more refined discourse of axiomatic set theory, TeX would not fail to deliver the requisite Hebrew aleph.

The markup completed, our well-trained toiler unleashed TeX, generating thereby foobar.dvi--in the speech of his tribe, a "device-independent file."

One task remained. Our punctiliously perspicacious TeXperson invoked a .dvi-to-.ps convertor, constructing a platform-independent deliverable, the Adobe PostScript file foobar.ps, ready even for a graphics bureau's highest-resolution laser printer and offset camera.

(b) Very different was our TeXnician's procedure by the late 1990s (that new world of Blair, of Putin, of Florida recounts fast and furious) with SGML-based automation ruling in short-lived triumph. Our editor, knowing SGML to be a tool for creating markup languages (a meta-markup language, celebrated in popular culture for creating HTML), astutely reached for the SGML-created language DocBook, and therein wrote foobar.sgml. In this strange new parlance - resembling HTML only in its overall angle-bracketed tag syntax, in its superfices - content was all, presentation nothing. In the DocBook language were, astoundingly, no codes for italics, none for large and medium-large type, none for horizontal and vertical rules. In their stead stood, in stern semantic purity, content-flagging codes for emphasis, for "sect1" or "sect2" headings, for "sidebar start" and "sidebar end." In the selection of typefaces and fonts for rendering the author's so punctilously coded semantic elements, DocBook spake not. Such authority was now vested instead in a free-standing stylesheet, authored in DSSSL by a typographer, keen of eye and sure in nuance, a rendering specialist living far from the madding particularity of individual documents. DSSSL, you ask? Here was a marvel unto our eyes, a new thing under the sun: a LISP-inspired language for specifying the typographic rendering of DocBook elements in document upon particular document, and rejoicing in independence from concrete typesetting technologies even as DocBook rejoiced! It was through the powerful partnership of DocBook and DSSSL that our TeXnician's task assumed its so stark simplicity: the briefest of bash-shell scripts did invoke, under Linux, the Jade typesetting engine. Jade, robotic servant unto two masters, the DocBook content and the DSSSL rendering directives, created a foobar.tex rich in "JadeTeX" macros, ready for bash-scripted conversions to foobar.dvi and foobar.ps. And lo, another marvel, worked by cunningly varied invocations of the so readily subservient Jade: our technician's fecund script now generated from foobar.sgml not foobar.ps alone, but also - albeit now without recourse to TeX - foobar.rtf (for Microsoft Word and its dull kin) and foobar.htm (for the HTML browser).

(c) A short-lived triumph for SGML, you ask? Necessarily so: for in the new millennium (those wretched days of Taliban, with anthrax at hand, briefcase-borne Hiroshimas impending) SGML deferred to its streamlined, machine-friendlier cousin, XML. And what, in this New World Order, of DSSSL? The human story was pregnant in pathos. Those same eminent programmers, those same distinguished practitioners of high cybernetic arts, who had brought DSSSL to be, did at the millennium's turn instead call forth XSL and XSLT. Their XSL and XSLT, they avowed, would make stylesheets platform-independent, even as DSSSL had, and yet would work within the very framework of XML, without DSSSL's troublesome exursions into the exotic Alonzo-Churchified lambda-calculus landscapes of LISP. And (so these high practitioners explained to the masses, schooled as those millions were in browser traditions) XSL and XSLT would bring to printing-press maturity the ideas more feebly incarnated in HTML 4.0 with Cascading Style Sheets. Thus it was that in these heady times. not authors alone, but cyberneticians, proved revolutionaries, firebrands. Would TeX achieve renewal and vitality in the upheaval? Or would lesser, weaker typesetting technologies be harnessed to XSL and XSLT to drive the finest PostScript presses of the dawning millennium?