Re: target language grammars.

A bit oblique to your question, but the document at


http://www.netfolder.com/DSSSL/DSSSL-markup-Rules.htm


discusses mapping sgml/xml "groves" (a trees representation of a
semantically marked up document) to output actions with dsssl.


Whether you use a grammar at all, represented as a parse tree or
otherwise, to map the output language semantics to the dsssl parser's
idea of the semantics of a given document is not really discussed.


At some point you have to explicitly tell the dsssl transformation
tool what elements of your output language map to what sort of dsssl
representations of parts of a document, using a dsssl stylesheet.


An on-screen representation is just another canvas (like a page in
LaTeX, postscript, pdf, pcl5, hpgl, etc). It may have it's own markup
language, but the semantic model isn't much different than that for
any other kind of rendered page.


TeX is already pretty good at abstracting this kind of difference away
with the dvi format, enabling dvips and all of it's brethren, but that
abstraction comes in rather late in the document translation process,
after you've already translated from an abstract semantic
representation of a document to some set of TeX macros interleaved
with the content.


SGML and XML move this process of semantic abstraction of document
elements upstream of any rendering markup, and that is probably a good
place to start on a problem of this nature.


Regards,


Clayton Weaver
<mailto: cgweav@aol.com>


"Everyone is ignorant, just about different things." Will Rogers

Return to the comp.compilers page.
Search the comp.compilers archives again.