We use URLs instead of HyTime, we will probably define our own style
sheets instead of DSSSL, so why shouldn't we define out own inclusion
mechanism outside of SGML, if we feel SGML is inadequate?
Which is inadequate: SGML or UAs that don't grok a prolog? I would
say the latter. Why invent another mechanism for external reference
when a suiable one exists and when in either case most current UAs
would have to be modified? Furthermore, why propose a mechanism that
requires changes to HTML when one exists that does not require such
a change?
I happen to believe that a UA should be able to understand some of
the prolog. It may behoove us to impose application conventions on
how much prolog must be understood. For example, we could stipulate
that a UA need not support anything but entity and notation declarations.
A UA would not require much more changes to support the latter than
any new inclusion mechanism. Besides, using entity declarations builds
on existing technology and doesn't require new technology.
Glenn