> I wager that it will be simpler in _most_ cases to use tags
> than PIs. Consider:
> <?DL (space-before: 2x font-size: 24pt font-weight: 'bold)>
> A Heading
> <?DL (space-before: 1x font-size: 12pt font-weight: 'medium
> start-indent: 3em end-indent: 3em)>
> First paragraph
> <H1>A Heading</h1>
> <p> First Paragraph
> The only time I could imagine seeing the former is from some
> RTF-to-SGML converter that just transliterated the presentation info.
This is a very tangible problem. As stuff gets imported from or
exported to HTML documents, we will need to be able to convert to
a variety of other formats, such as Word/RTF, Latex, Mif/mml, etc.
If the source info includes physical formating, then the converter
will have to have heuristics to know that "a new line, a small indent,
followed by flush or right-ragged non-indented text" is a "(indented)
paragraph", and hence to handle it as such.
This is messy. Easier just to convert from "paragraph" in one
high-level format to "paragraph" into "paragraph" in another format.
The only reason that RTF has \pard is because it saves you about 40
bytes of having to write out the equivalent notation in raw
formatting commands... (well, almost ;-)