Re: Possible use of <INSERT> - serious inquiry (fwd)

Marc Salomon (
Thu, 21 Mar 1996 11:35:56 -0800

From: "Marc Salomon" <>
Message-Id: <>
Date: Thu, 21 Mar 1996 11:35:56 -0800
In-Reply-To: MegaZone <>
Subject: Re: Possible use of <INSERT> - serious inquiry (fwd)

MegaZone <>:
|I'm saying lets change what 'valid HTML' means, *if* that would provide
|increased functionality on the web.

Functionality on the web does not exist along a single axis in a single
dimension.  Decisions made to facilitate rendering on a GUI client can take
their toll when it comes time for other user agents that have no interest in
rendering an object on the screen want to proces a same document.  Let's note
the difference between browser and user agent.

Data repurposing is the biggest win of SGML.  Most content created on the web
today (particularly using the HTML distensions de-jour) will be obsolete in 3-6
months in any case, so there needs to be a system that can include both
non-parsing ephemeral fluff as well as high-value added persistent, reusable
information.  One of the first requirements for a WWW browser from the early
CERN documentation was that they be forgiving when rendering imperfect

Great care has been taken to craft a valid ISO-8879 DTD for HTML.  Aany
applications are available that depend on the SGMLness of HTML.  There is a
middle way where the SGML purists can accept mixing structure and presentation
in a document in an SGML compliant manner, avoiding the proliferation of
special tags and attributes describing style.  But to trash SGML for the
convenience of GUI user agents trashes years of hard work creating the first
large-scale, open digital information interchange infrastructure.

On the insert question, it is now possible to create nonconforming composite
documents in several ways, from vi up to an expensive authoring tool to various
server-side techniques and even >gasp< SGML entities.  Authors need to take
some responsibility for ensuring that the documents they cook up are valid if
they want their information to be reusable--this cannot be mandated or coerced.
 Most information on the web has no need to be crafted for reusability.  But
the fact that some content providers don't need this functionality is not a
valid argument for constricting the functionality of HTML and its successors
for those of us who do.

An otherwise conforming HTML document with INSERT tags is still a conforming
SGML document.  I would argue that running that conforming document through a
preprocessor that understands how to resolve the HTML <INSERT> tag, and then
piping the resulting compound document into an ISO-8879 parser to check for
conformance breaks SGML.

All I'm saying is let's not wear blinders when it comes to the wins of SGML.


// Marc Salomon - Software Engineer - Innovative Software Systems Group     \\ 
\\ Library and Center for Knowledge Management - UC, San Francisco          // 
// phone :  415.476.9541 - e-mail : - fax: 415.476.4653   \\