Re: Tag Soup (was: FW: XHTML)

On Sun, 5 Dec 1999, Daniel Hiester wrote:

> Well, then, if Tag Soup will continue to be a problem 

Tag Soup per se is not a problem.  It's merely inadequate as a paradigm.
This was realized 30 years ago - yes, that's how old the concept of
Generalized Markup is.  See

  http://www.sgmlsource.com/history/roots.htm

The Mosaic pheonomenon was a gigantic leap backwards, but human vanity
prevents acknowledging that.  (Too many people are on record singing their
hosannas to the Mosaic spawn - much better, it would seem, to find some
face-saving way to describe Tag Soup as "advanced technology" or "vendor
driven progress" or somesuch.)

> then why not just make tag soup incompliant w/ XHTML?

Well, it's important to realize that the term 'HTML' has been shanghaied.
HTML *is* Tag Soup.  Tag Soup *is* HTML.  We need a different name (and we
need to drop the 'L'), and to make the incompatibility explicit, we need a
spec for Tag Soup.

> When I wrote Tag Soup, I never used a Doctype Declaration... almost nobody
> did back then. (and at that, it wasn't really tag soup, it was "kinda" tag
> soup) 

> If a Document is going to be well-formed, then it must include a
> doctype declaration. 

Not so.

> If it isn't well-formed, then it shouldn't include a doctype
> declaration.

This should be: include a doctype declaration only if you need it for
SGML/XML conformance (i.e. you intend that a SGML-aware system take the
document seriously.)

> Let tag soup continue to exist; but if tag soup authors put doctype
> declarations into the documents, the parser should just totally ruin
> it.

This won't work.  Frontpage routinely inserts bogus doctype declarations
in its extrusions.  So do HotDog and any number of equally bogotic "HTML
editors".  Netscape Composer in its RTFM-challenged excellence elects to
convert everything to lower case too.  The lesson here is that familiarity
breeds comtempt.  

But there is a way out: the Content-type header.  'text/html' should be
the Media Type for Tag Soup.


Arjun

Received on Monday, 6 December 1999 17:11:09 UTC