Except microformats are a bit of Hytime:  the arch forms.  And despite
the almost mythical original goal (it would be hard to get agreement on
what that was), XML has been applied very much where it was said it
wouldn't be: messaging.  I proposed that on comp.text.sgml and was
quickly shot down.  The MID proposed XUL/XAML and Yuri looked at me
across breakfast and said quite seriously, "No one will ever do that."
So much for the vision of XML.  In the end, the groups synthesized the
common experience and it was applied as applicable.

Hytime baked for too long.  It starts out with the same linking as HTML
but spins out of control as ever more stuff is added.   It is the
classic "complexity doesn't scale or sell" story, so bits and pieces get
picked up and redone as independent works.  (As I recall, that was
predicted.)  Timing is everything; listening is everything else.

Interesting: given all of the various projects over the years, if one
started with SGML again looking at the various projects, what would the
subset be today?  Or would it be a superset now (SGML almost had a
binary but the roof blew in just as that was happening)?

len


From: Robin Berjon [mailto:robin.berjon@e...] 

And microformats, Ye Aulde HTML, etc.. XML picked a subset of SGML that
fit a specific goal. I wonder (completely idly) if there wouldn't be
value in extracting the minimisation features from SGML to create some
form of "XML (Infoset) Extraction" spec, similar to what is being done
with GRRRDDL?