Re: Cougar DTD: Do not use CDATA declared content for SCRIPT

Paul Prescod (papresco@calum.csclub.uwaterloo.ca)
Sat, 27 Jul 1996 09:21:27 -0400 (EDT)


From: Paul Prescod <papresco@calum.csclub.uwaterloo.ca>
Message-Id: <199607271321.JAA20600@calum.csclub.uwaterloo.ca>
Subject: Re: Cougar DTD: Do not use CDATA declared content for SCRIPT
To: davidp@earthlink.net (David Perrell)
Date: Sat, 27 Jul 1996 09:21:27 -0400 (EDT)
Cc: www-html@w3.org, ack@ebt.com
In-Reply-To: <199607270217.TAA17801@armenia.it.earthlink.net> from "David Perrell" at Jul 26, 96 07:16:57 pm

> Really? An HTML document is marked up to instruct the
> browser/formatter/renderer what to do with the content. 

An HTML document is marked up to instruct the browser/formatter/render/
editor/search engine/natural langauge translation tool/text analysis tool/
VR renderer/speech translator/table of contents generator...

The whole point is that you do not and cannot know what tools will work
with your HTML document. If you are designing with some finite set of
tools in mind, you should redesign.

> Instead of expounding on the true nature of SGML and the appropriate
> place for style, why not just address my premise, which is that if
> inline scripts are to be "legal" in an HTML document, they should not
> be enclosed in such inelegant and unintuitive constructs as
> <![CDATA[this is <foo> text that &bar; should not get parsed]]>?

If inline scripts are to be "legal" in an HTML document they should be 
enclosed in a construct that gives the HTML authors at least a fifty/fifty
chance of making correct documents. The CDATA construct is the only way.
Alternatively, we could just make this illegal, instead of repeating the
mistakes of other badly designed DTDs.

 Paul Prescod