- From: Dan Connolly <connolly@w3.org>
- Date: Mon, 13 Dec 1999 14:38:23 -0600
- To: spec-prod@w3.org
The XML spec DTD http://www.w3.org/XML/#xmlspec is pretty cool in a lot of ways, but it seems to me that it's arbitrarily different from HTML in expensive ways: you *have* to use a batch process to convert from source form to preview/display/delivery form. Back in 1995, I wrote: Ideal Solution Source format: HTML dialect use a strict HTML dialect with: tables, class=abstract, possibly math. Document Manipulation API: java interface -- http://www.w3.org/MarkUp/SGML/spec-mgmt and in May 1997, I did a little hacking http://www.w3.org/XML/9705/hacking on a (psuedo-)XML parser and some tools to convert an HTML dialect to lout for typesetting. The example I used was: http://www.w3.org/Architecture/NOTE-ioh-arch i.e. source: http://www.w3.org/Architecture/NOTE-ioh-arch.html output of lout conversion: http://www.w3.org/Architecture/NOTE-ioh-arch.lt postscript output from lout: http://www.w3.org/Architecture/NOTE-ioh-arch.ps along with http://www.w3.org/MarkUp/9705/report.dtd which "refines" HTML: http://www.w3.org/MarkUp/9705/html.dtd to add report titlepage stuff (abstract, ...) a section/subsection structure, bibliography stuff, etc. (note that I used an extension to SGML marked section syntax for modules) Now that we have XHTML (nearly) and DOM and XSLT, I hope to revisit this idea and finish the code, but I haven't managed to find time, so I'm sending this message to see if anybody else is interested/motivated. For example, my conversion program was rules-based, and I hope that it converts to XSLT straightforwardly: http://www.w3.org/XML/9705/report2lout.py The idea is to use on XHTML dialect for editing *and* delivery. It has some redundancy that is (or at least: could be) managed by machine; for example for quotations: <q><c>"</>Hyperdocument<c>"</></q> If you're using lynx or Mosaic 2.0 or something, you just get "Hyperdocuments" but you can also do a stylesheet that adds real printer's quotes and supprsesses c as a child of q. If you don't want to add these <c>"</c> things by hand, you can do it automatically using DOM scripts or XSLT. Same goes for all sorts of generated text: tables of contents, cross references, indexes, ... . Not to mention specialized notations like grammars and such. The idea is: you generate those by machine, but you don't treat the results as junk to be thrown away; you fold it back into the source. -- Dan Connolly http://www.w3.org/People/Connolly/
Received on Monday, 13 December 1999 15:38:31 UTC