[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]
xanadu://path or http://xanadu-path (RE: "xurl?)
- To: xanadu@xxxxxxxxxxxxx
- Subject: xanadu://path or http://xanadu-path (RE: "xurl?)
- From: ____Textpert Alert____ <ianf@xxxxxxxxx>
- Date: Wed, 15 Nov 1995 19:00:32 +0100
- Reply-to: xanadu@xxxxxxxxxxxxx
Writes Steve James <sjames@xxxxxxxxxxxxxxx>
Ted>> But two of my collaborators, Roger Gregory and Sam Epstein,
Ted>> think that transcopyright protocol should be recognizably
Ted>> different at the top level, eg
Ted>> xu://whatever@xxxxxxxxxx
> To put my $0.02 in, This seems reasonable.
To begin with taxonomy, that "top level" has a name, and it
is "scheme," rather than "protocol." The URL/URN/URI syntax
decomposes into
scheme://path
parts, all nicely explained in
http://www.w3.org/pub/www/doc/url-spec.txt ; may have gone cold
While instinctively agreeing with Tim Berners-Lee that transcopy-
righted material could well be served using the HTTP protocol for
the transport layer, but with proprietary negotiation, validation,
fragment-assembly and billing mechanisms incorporated along the
way (which need to be there no matter what transport methods have
been chosen), I have come to realize that having a dedicated
protocol called "xanadu," ie. having URIs begin with that word
would offer clear advantage for the users and publishers alike.
(URIs not URLs, because they won't be simple URLs anymore; see
http://www.w3.org/pub/www/doc/uri-spec.txt ; may have gone cold)
For one, no publisher could claim to have accidentially linked
in a fragment for which his readers subsequently end up paying.
Since any seamless, transparent transcopyright mechanism also
means that end-users lose a degree of control over what they are
served, we need to ensure that there is ample room for filtering
out potentially-unwanted bits, also based on any particular
scheme ALONE.
Once the WWW becomes a charge-by-the-byte medium, browsers will
need to be able to recognize, and _filter out_ both the given
AND reg-expressed schemes, servers and documents (ie. a user
instructs his browser to ignore all xanadu://xu.whitepower.com/*
fragments, and can forget about ever seeing anything from there).
For the sake of this argument alone I assume that Xanadu-servers
would be implemented as a superclass of HTTP-servers, ie. capable
of dealing also with non-Xanadu browser requests, though perhaps
only delivering some boilerplate text ("This 123456 byte long
fragment of _What I Did On My Summer Vacation_ by A. Pupil can
only be seen by Xanadu-capable browers. Get one from [here]")
in lieu of the correct fragment in such cases.
Observe, however, that I advocate the scheme called "xanadu:", not
"xu:", which is much more vacuous and ambiguous than that. There
already is a valid URL scheme called "prospero", so "xanadu"
wouldn't be out of place in such company. Ergo:
xanadu://xu.w3.org/hypertext/WWW/Xanadu.html#transcopyright
__Ian