Freezing the HTML spec Re: Comments in HTML ?

Dan Connolly (connolly@pixel.convex.com)
Thu, 19 Nov 92 12:49:51 CST 

Message-Id: <9211191849.AA13833@pixel.convex.com>
To: timbl@nxoc01.cern.ch
Cc: www-talk@nxoc01.cern.ch
Subject: Freezing the HTML spec Re: Comments in HTML ? 
In-Reply-To: Your message of "Thu, 19 Nov 92 18:27:46 +0100."
             <9211191727.AA00350@www3.cern.ch> 
Date: Thu, 19 Nov 92 12:49:51 CST
From: Dan Connolly <connolly@pixel.convex.com>


What is the plan for registering HTML with the IETF? The HTML
spec seems to be somewhat in flux. Tim talks a lot about
"HTML futures." There are currently a lot of open issues.
Eventually we have to "shoot the engineers and ship it,"
that is freeze the spec and hand it to the IETF.

I suggest the following steps:

1. Establish a charter for the HTML spec to guide us in
resolving issues -- a sort of requirements spec. Here's
my working list:

	* Establish a well-defined relationship between
	HTML and SGML. Don't include anything in the HTML
	spec that conflicts with the SGML spec.

	* Use the LineMode browser as a reference implementation.
	Try not to include features that www doesn't grok.

	* Support common structural markup: headers, lists, etc.

	* Support W3 addressing syntax.

	* Support links from ranges of text to other documents
	and anchors in other documents.

	* Keep the structure somewhat flat for interchange with
	PC style word processors and other style-per-paragraph
	oriented processors.

2. Establish an HTML spec document, some informative appendixes,
and some examples/test cases. I suggest:

spec: ftp://info.cern.ch/pub/www/doc/html.txt
	from http://info.cern.ch/hypertext/WWW/MarkUp/MarkUp.html
	and its neighbors "Text.html" "Tags.html" "Entities.html"
	and "HTML.dtd"

infomative appendices: "HTMLConstraints.html" "Future.html"

examples: "../Test/test.html"

3. Revise the spec so that it's internally consistent. Right
now, there are some glitches. And the current method of
sending suggestions to Tim and hoping he finds time to make
the edits is no good. Hmmm... we definitely need a CSCW
strategy for group-editing of documents.

4. Register the spec with the IANA or IETF or whatever.

Meanwhile, I think it's pretty important to fix the NeXT editor
and all the files on info.cern.ch. Folks are using that as a
reference, and perpetuating HTML that conflicts with the SGML
standard.

Back to the question at hand...


>Begin forwarded message:
>
>> Date: Thu, 19 Nov 1992 13:14:33 +0000 (GMT)
>> From: Kevin Hoadley <K.Hoadley@directory.rl.ac.uk>
>
>
>> Is there any way to include comments in html documents ? In sgml ?
>
>In SGML, yes there is a -- comment start and a -- comment end.
>Someone corerct me if I am wrong but I understand that once within
>a tag one can use these ... right? Or does it apply to DTDs only?

The <!-- blah blah blah --> construct is only recognized in
the DTD. There's a thing called a processing instruction that
looks like <? blah blah blah> which is recognized in the
instance. The Linemode browser seems to grok.

I think we should mention this construct in the standard, since
we're technically not allowed to disallow it or process it incorrectly.

There's another thing called a marked section that looks like
<![status [ stuff ]]> where status is one of...

IGNORE means ignore this stuff, except recognize nested marked sections.
INCLUDE treat stuff normally
CDATA 	treat stuff as data. Recognize only ]]>
RCDATA 	treat stuff as data. Recognize entity reference (&foo;) and ]]>
TEMP	(why is this in the SGML standard?)

The <![CDATA [ ... ]]> should probably be supported for plaintext
sections somehow, but I can't figure out a scheme that harmonizes
with the linemode implementation.

There's also some character classes MSICHAR, MSOCHAR, MSSCHAR
for markup suppression. Markup between MSICHAR and MSOCHAR is
ignored, and the next character after MSSCHAR is treated as
data rather than markup. Unfortunately, these classes don't
have any members in the reference concrete syntax, and they're
probably not supported by many parsers.

SGML is a mess!

>> Does any of the existing WWW code support comments ?
>
>As it happens, the current library supports them, so the line mdoe
>browser and anything else based on the library does. But it has
>been left out of the doc and so will probably me missing from other browers.
>

Try the <? foo > construct.

Dan