On 10/20/2012 11:43 PM, don kretz wrote: [snip]
I think no one is advocating any form of text that serves both purposes simultaneously,
No, that is precisely what I am advocating. If you carefully segregate the semantics/structure of a text from its presentational elements, and have a technology that allows these two elements to be merged at the time of use, it is possible to mark up a text so that it can be used directly by an end user as well as being the source for transformation into other markups. As it turns out, we do have a technology that allows semantics and presentation to be merged at the time of use: XML + CSS. And most available user agents, such as HTML browsers and ePub-based readers, are capable of performing this merging (MobiPocket and first generation Kindle's are notable exceptions to this rule). HTML is not the only markup language that can serve this function. In the past I have created a CSS file that could be attached to a TEI file and which resulted in browser rendering equivalent to HTML + CSS. HTML simply has the advantage of a set of predefined styles that are used as basic styles which can then be overridden by internal or external style sheets. In the absence of style sheets HTML can provide a mostly acceptable presentation, whereas without explicit style sheets other XML vocabularies have no presentation whatsoever. (Although, this does not preclude someone from creating a TEI user agent that /does/ have predefined styles for TEI elements. I think I may be able to force Cool Reader to do just that; yet more experimentation is needed....) So yes, I want to define a uniform set of semantic inflection (class attributes) that can be applied to HTML files that will allow them to be used as a master file that can be transformed into any other markup language, used in combination with a CSS file to provide rich presentation in HTML-based user agents, and which can degrade gracefully in HTML-based user agents when no CSS styles are present.