> I like the general idea of coming up with a well defined set of
> extensions for the "#xxxxx" part of uniform resource identifiers, for
> pointing to not only HTML tags, search strings in text/* documents,
> image coordinates, etc.
I'll second that.
The WN server does a server-side interpretation of URL parameters,
/dir/foo.txt;lines=10-20
^
would be interpreted as a requiest for lines 10-20 of "foo.txt".
(See <URL:http://hopf.math.nwu.edu/docs/range.html>).
Perhaps a similar syntax would be appropriate for client-side
interpretation of URL fragment identifiers:
http://foo.com/dir/foo.txt#lines=10-20
^
The presence or absense of an '=' sign in the fragment would
determine whether the fragment is an extended locator
or an anchor name (or element ID in HTML 3).
Whatever syntax is chosen, it should be extensible.
There are lots of possibilities:
foo.txt#lines=10-20 -- highlight lines 10 through 20
foo.html#textsearch=radioactivity -- first (or all) occurrences of phrase
foo.html#teipath=P+3+EM+1 -- first <EM> in 3d <P>
foo.html#treeloc=1,5,3 -- HyTime treeloc location
foo.html#nameloc=anchor1 -- same as foo.html#anchor1
Text searches and TEI-style paths would be most useful.
--Joe English