> Joe writes:
> >If style information is to be embedded in the document body,
> I feel it should at least be synchronous with the element structure.
> Changing style sheets "midstream" defeats this; they would
> act more like PIs.
>
> and I wonder how the scope of such <style> info is to be determined.
> Hakon, is this part of your proposal? if so, could you summarize?
Allowing style information interleaved with HTML content was part of my
first style sheet proposal. I dropped it after doing a little thinking
of the issues involved:
- The scope and inheritance of the inline style specification
become less than intuitive. Should the style hints apply only to
the element where they are specified, or be taken along to the
next one? In either case, you will end up sending more information
than with the class mechanism of HTML3 combined with the cascading
style sheet proposal.
- By making order significant in the style specification, we are opening
up for ambiguities. E.g. if one refers to a value (e.g "H2: font.size =
H1:font.size * 0.8") that is being changed in a later style
specification, should you go back and update the first one? Of
course, one could make it illegal to refer to other values, but
then you're also taking away some of the expressive power of the
language.
- The style sheet (i.e. the collection of style hints) can no longer
be applied to more than one document. It will e.g. be impossible
to tell your browser "make the current style sheet my personal
favorite" since the style sheet is inseparable from the document.
- If all documents are to include a style description, the required
bandwidth of the web will increase significantly. By using LINK to
refer to a style, one can reduce bandwidth (by caching the most
popular styles) as well as latency (fetch the style after the content).
- The fundamental motivation behind putting style on the web is
aesthetics. By putting style information all over the document,
the source of the HTML source will look messy. Like Netscape's
oversize caps scheme does today ("<FONT SIZE=+3>C</FONT>ONTROLLING").
Enough people will have to read and edit the source to make this a
valid argument.
Cheers,
-h&kon
Hakon W Lie, WWW project CERN, CH-1211 Geneva 23
http://www.w3.org/hypertext/WWW/People/howcome/