Re: MIME vs. HyTime

Dan Connolly ([email protected])
Mon, 30 Nov 92 22:02:29 CST


>I claim no expertise on HyTime, but I believe the intent
>is to allow multimedia data in an SGML DTD.

Good point. I don't know as much about HyTime as I should
either. All I know is that it's complex and obscure enough
that I don't have a working knowledge of it after two
years of sniffing around the internet.

> Given Dan's
>current efforts to make HTML a true DTD, it seems like
>HyTime tags might be an easier addition than incorporating
>HTML as a MIME datatype.

I'm almost sure that's not true. Incorporating HTML as
a MIME datatype is as easy as sending an email message
to the IETF. Now the interesting stuff: incorporating
the nifty features of MIME into WWW is anther story.
But I still think it's several orders of magnitude
easier than implementing a HyTime engine.

> Has anyone looked at HyTime vs.
>MIME?

The relavence of HyTime to WWW isn't so much in the
realm of data formats (where MIME is key), as in
hyperlink semantics and addressing schemes.

HyTime architectural forms have immense expressive
power, but I gather they're pretty heavy to implement.
They do stuff like:

<LINK target=loc1>click here to see a film</LINK>

<FILM HyTime=FCS ID=loc1>REEL 100<START>100.23sec<STOP>134.56sec</FILM>

I don't have a firm grasp of how it all works, but most of
the link mechanisms involve cooking up some element that
describes referent data, and then using the id of that
element elsewhere in the document.

It would mean turning
<A HREF="http://info.cern.ch/hypertext/WWW/TheProject.html>Click here.</a>

into

<A HREF=loc2>Click Here.</a>

<RESOURCE ID=loc2><SCHEME>http<HOST>info.cern.ch<PATH>
<ROOT>hypertext<COMPONENT>WWW<COMPONENT>TheProject.html</RESOURCE>

or something like that.

> I guess I'm wondering about this for the mail world
>as well, since there is already a great deal of commercial
>interest in SGML-ifying all documents.

The purpose of SGML is interchange. It's a pretty painful
investment unless you want to exchange documents among
diverse systems at the source level. The WWW project
is paying the price, and I think it's working well.

So far, WWW is just for text, where SGML is sufficiently
epressive. SGML has hooks for the kind of multimedia
that WWW is about. HyTime is overkill for graphics and simple sounds,
IMHO.

If WWW ever involves complex multimedia documents, especially
documents where timing and event ordering is significant,
we might want to take another look at HyTime.

Dan