Re: HTML is not SMGL

Dan Connolly (connolly@pixel.convex.com)
Sun, 07 Jun 92 22:29:44 CDT


Message-Id: <9206080329.AA28300@pixel.convex.com>
To: Edward Vielmetti <emv@msen.com>
Cc: jfg@dxcern.cern.ch (Jean Francois Groff), www-talk@nxoc01.cern.ch
Subject: Re: HTML is not SMGL 
In-Reply-To: Your message of "Sun, 07 Jun 92 20:26:48 EDT."
             <m0luXZf-0009YoC@nigel.msen.com> 
Date: Sun, 07 Jun 92 22:29:44 CDT
From: Dan Connolly <connolly@pixel.convex.com>


>The UDI vs. MIME argument is a non-arguement.  MIME is sufficiently
>flexible that if you construct an appropriate Content-type and define
>its semantics appropriately it will accept UDI's and work accordingly.
>"Simple matter of programming" :).
>
>Explicit "attribute=value" tags are more flexible than the W3 approach
>to turn the entire document ID into a big long string.  I guess it 
>depends on whether you believe you are dealing with a big database
>or a big file system.  Both approaches have their place.  Again as
>a simplified case you have "udi=//host:port/path" as a MIME identifier
>and all is well.
>
The problems is that the syntax of a UDI doesn't fit into the syntax
of a MIME parameter (or an SGML attribute value...) because a UDI
might be arbitrarily long, and it cannot contain any whitespace (so
it can't be split across lines).

So these 200+ character UDI's for WAIS documents can't be
mailed around safely (even SGML has limits on the length of an
attribute value).

Heck, my WWW client (perhaps it's not the latest version, but still...)
can't even retrieve wais documents due to these problems.

Dan