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

Paul Prescod (papresco@calum.csclub.uwaterloo.ca)
Mon, 29 Jul 1996 17:29:45 -0400


Message-Id: <1.5.4.32.19960729212945.00c12f70@csclub.uwaterloo.ca>
Date: Mon, 29 Jul 1996 17:29:45 -0400
To: <www-html@w3.org>
From: Paul Prescod <papresco@calum.csclub.uwaterloo.ca>
Subject: Re: Cougar DTD: Do not use CDATA declared content for SCRIPT

At 01:33 PM 7/29/96 -0700, David Perrell wrote:
>Albert Lunde wrote:
>> One either has to stay within the rules of SGML _or_ produce
>> a formal description of how to go outside them. Without
>> a formalism, it's just a pile of hacks.
>
>The formalism itself should not be a hack. If a script cannot currently
>be legal SGML content, then either a content type should be devised
>that works or SCRIPT should not be formalized in its current form at
>all.

Since SGML was not designed to be a file format for containing random data
types, any formalism that allows inline scripts is going to be a hack.
Languages designed to contain random data types (like MIME) either have a
construct for declaring the "end of data string" or embed the size of the
data at the top. Even if we could extend SGML to allow one of these, the
chances of authors going to the trouble is remote. They would either have to
count their bytes or invent a unique "end of data string" for each script.
Most likely they would all use the same "end of data string" (<\SCRIPT>)
that they see on Netscape's site, even when their script contains the string
<\SCRIPT>.

Since there is no way to standardize inline scripts that will result in
reliable interoperability, they should be disallowed.

 Paul Prescod