Re: Separating Code and Data

Simon North (north@Synopsys.COM)
Wed, 18 Feb 1998 14:32:13 +0001


> 2. I'm not sure what's static or dynamic about XML/SMIL or HyTime
> (also an SGML application, btw). The actual event logs can be static
> or generated dynamcially; the logical timelines seem to be equally
> expressive. But, I'll confess to never having properly understood
> HyTime. Perhaps you can elucidate...

I'm risking my neck here ... I've attended HyTime courses, I'm even
on a HyTime-related standards commission, but my knowledge of it is
far from inclusive. But, I'll have a go.

HyTime is to SGML as SGML is to XML. Glib statement, but HyTime is
officially an SGML application while for all practical purposes it is
a superset of SGML. HyTime adds 'architectural forms' to an SGML DTD.
Architectural forms are a direct analog of classes. This is the core
of HyTime, but it brings with it a series of modules that can be used
to locate and address any kind of data. XML and SMIL use a very, very
little of HyTime's functionality by adopting parts of its linking
mechanisms (XML) and addressing/locating mechanisms (SMIL).

As an example of such a module, in the distant past HyTime had an
SQL query module, HyQ; this became SDQL.

Check out "Managing SGML Architectures and Object Models with Groves"
at http://www.isogen.com/papers/grovemng/grovemng.html

The definitive source is, of course, Eliot Kimber's page (Dr. Macro)
at http://www.drmacro.com/hyprlink/index.htm -- via this page you can
also access (after online registration) a draft copy of his
forthcoming book on HyTime.

Simon.