Re: A Proposal from the Disabled Access Community

BruceLeban@akimbo.com
Fri, 18 Apr 1997 23:41:17 -0400 (EDT)


From: BruceLeban@akimbo.com
Date: Fri, 18 Apr 1997 23:41:17 -0400 (EDT)
Message-Id: <199704190341.XAA13819@mail.internet.com>
To: www-html@w3.org
Subject: Re: A Proposal from the Disabled Access Community


>From:	progman@ecst.csuchico.edu (James Shattuck)
>In other words citations would not be 
>limited to anchors named by the author, but an URL could go precisely 
>to a location defined in the cited document by a third party.
>scheme://path.to.server/path/to/file.typ?where-it-says=I%20told%20you%20so 

I don't think that changing the URL syntax or overloading ? is necessary 
for this purpose, especially since such a change would cause 
compatibility problems. However I note that most browsers ignore a 
nonexistent fragment id so:

    http://www.bogus/etc.html#foobar

would position to the beginning of the document if there is no foobar 
anchor. Overloading # for this use would thus provide graceful fallback 
in browsers that don't support this feature and furthermore it's 
consistent with the general idea of a fragment reference. 

Browsers could be modified to search for text if they can't find the 
anchor. If they can't find the exact phrase they could look for a 
paragraph containing most of the words or whatever. This would allow the 
reference to work as well as possible if the document changes.

If this was really going to happen it might be a good idea to do a bit 
more than just searching for a string. E.g., allowing specification of a 
range of a document like:

    http://www.bogus/etc.html#foo#bar

which could refer to the portion of the document from the first instance 
of "foo" to the next instance of "bar". (Not very useful for a graphical 
display which would just scroll foo to the top, but this would be very 
useful for a speech-based browser.) Another idea would be allowing 
reference to the nth instance of a string, something like:

    http://www.bogus/etc.html#foobar##3
for the third or for the last instance:
    http://www.bogus/etc.html#foobar##-1

Interestingly, I couldn't determine whether or not it's generally 
believed that there must be exactly one # or not. E.g., RFC 1738 
acknowledges the existence of fragments, but other than saying that they 
appear after a # it says nothing else (although it references a URL using 
one so clearly they know how it works :-).

    --- Bruce Leban
    Akimbo Systems
    http://www.akimbo.com/globetrotter
    Publish on the web without learning HTML! (Really.)