Something Tony and I were recently discussing is the possibility of making a
"standard" preprocessor for HTML documents. This way, you would have regular
HTML documents (which would not be parsed, just sent by the server), and you
would have a new type like HTML-macroized. When the client requests a
document of this type, the server would run the pre-processor on the
HTML-macroized document, which would then spit out an HTML document to
return to the client.
This way, it would be "just another document conversion" to the server, and
if we agree on a standard preprocessing format, it will work on all servers
that support it. It also does not add tags to HTML which really don't belong
in HTML itself.
What do you think?
--Rob