Re: Include pieces of code in an HTML file (fwd)

On Fri, 1 Oct 1999, Russell Steven Shawn O'Connor wrote:

> On Fri, 1 Oct 1999, Arjun Ray wrote:
> 
> > The book was closed on this a long time ago.  See
> > 
> > <URL:http://www.deja.com/=dnc/getdoc.xp?AN=259842602>
> > <URL:http://www.deja.com/=dnc/getdoc.xp?AN=259906443>
> 
> This just says that adding entities is illegal in HTML.  I agree with
> that.  

No, it says that internal subsets are illegal in HTML.  Never mind that,
strictly speaking, it's *external* subsets that are redundant!  That is,

 <!DOCTYPE HTML PUBLIC '-//W3C//DTD HTML 4.0 Transitional//EN' >

is just "syntactic sugar" for something like this

 <!DOCTYPE HTML [
  <!ENTITY % dtd PUBLIC '-//W3C//DTD HTML 4.0 Transitional//EN' >
  %dtd;
  ]>

(See p. 402 in _The SGML Handbook_ for the clarification that indeed
nothing more than just syntactic convenience is involved.  Dr. Goldfarb
also proposed eliminating the external subset syntax for XML, btw.)

Nevertheless, the HTML specs "say" that the first form is legal but the
second isn't.  What the verbiage ("to facilitate experimentation and
interoperability") manages to hide very well is the *motivating* fact: 
that everyone's darling wowsers manage to stumble their way past the
"legal" form but barf on the canonical alternative.  Since they're going
to ignore the stuff anyway, might as well make them look good doing so. 

> I said that I think I the specs should be changed to allow this
> construct.

I agree, but I seriously doubt that this will happen, ever.

> > Because Netploder doesn't support it.  The web is what Netploder supports.
> 
> Netscape and Explorer don't support the Q element as far as I know, yet
> the element was added to the specifications.

Element types are a different matter.  The root issue has always been the
lack of support for the Reference Concrete syntax in the Mosaic-spawn. 


Arjun

Received on Friday, 1 October 1999 21:05:41 UTC