- From: William F. Hammond <hammond@csc.albany.edu>
- Date: Sat, 31 Oct 1998 09:07:36 -0500 (EST)
- To: hutch@psfc.mit.edu, jjramsey_6x9eq42@yahoo.com
- Cc: emj@cnsibm.albany.edu, www-math@w3.org
This discussion is about "authoring", which remains largely without provision, and explicitly so to date, in the work of the W3C Math WG. MathML is browser fodder. I would like to consider a simple example, namely the classical TeX system-building benchmark "story.tex" in the standard web2c-TeX distribution from Stanford, in relation to the question of how it might be translated into HTML. As I think about this, I ask the question why HTML rather than the SGML language known as "DocBook"? It seems to be widely accepted that HTML is more of a "sink" than a "source". The "sink" v. "source" thing is, of course, relative. DVI is a sink relative to TeX, while DVI is a source relative to PostScript. (And, yes, both DVI and PostScript are formally specified languages.) "DocBook", in which I am not fluent, is said to be a source relative to languages in which I might choose to author as source, including both LaTeX and HTML. In fact, a visible demonstration of this is available if one acquires the archive (from CPAN, the Comprehensive Perl Archive Network) for David Megginson's Perl library SGMLSpm. (Megginson has other newer things for XML at "www.megginson.com" that may wind up being useful for us.) So why not try to set "story.tex" in "DocBook"? Beyond that, why not try to set it in a DTD that is a bit closer to the world of LaTeX that I know? I come up with this: <!DOCTYPE article SYSTEM "gellmu.dtd"> <article> <preamble> <title>A Short Story</title> <author>A. U. Thor</author> </preamble><body> <hrule> <parb> Once upon a time, in a distant galaxy called <umlau>O</umlau><umlau>o</umlau><cedil>c</cedil>, there lived a computer named R.<nbs/J. Drofnats<eos/ <parb> Mr.<nbs/Drofnats<pdash/or <quophrase>R. J.,</quophrase> as he preferred to be called<pdash/<nul/ was happiest when he was at work typesetting beautiful documents<eos/ <hrule> </body> </article> Perhaps I should comment on three of these tags: "<eos/", "<nbs/", and "<nul/". These are empty SGML elements, like HTML's "<br>". They *could* be tagged, and with more standard SGML usage, *would* be tagged as "<eos>", "<nbs>", and "<nul>". Moreover, in XML, they *must* be tagged as "<eos/>", "<nbs/>", and "<nul/>". My usage is enabled by tweeking the SGML DECLARATION [I began with that for HTML 2.0, found in Internet RFC 1866, section 9.5, rather than manually copy the reference declaration from Charles Goldfarb's "SGML Handbook".] to enable "SHORTTAG". Other features not in the default reference declaration (the foundation of SGML and SGML DTD syntax) were already there. I gain thereby the advantage of having my parser know about the emptiness of these tags before I put them in my DTD, which is useful if I want to use sgmlspm to make a format chart of tag logic. The price is that the character "/" needs to be made special (in one of several possible ways). I say all of this just in case the Math WG is still working with an outside SGML consultant. "eos" is a formal end-of-sentence mark. "nbs" is non-breaking space like "~" in LaTeX and " " in HTML, and "nul" is comment residue. Comment? What comment? Due to my persistent LaTeX habit, I actually set this as follows for simple transliteration. But it's GELLMU, not LaTeX. After all, "\hrule" is not (Lamport) LaTeX, and "\documenttype" is neither LaTeX nor TeX. \documenttype{article} \title{A Short Story} \author{A. U. Thor} \begin{document} \hrule Once upon a time, in a distant galaxy called \umlau{O}\umlau{o}\cedil{c}, there lived a computer named R.~J. Drofnats. Mr.~Drofnats---or \quophrase{R. J.,} as he preferred to be called---% error has been fixed! was happiest when he was at work typesetting beautiful documents. \hrule \end{document} Oh, by the way, didn't Thor used to be reachable as "author@cs.:o:o,c" ? ;-----) -- Bill
Received on Saturday, 31 October 1998 09:07:41 UTC