Roy T. Fielding fielding@ebuilt.com
Wed, 13 Jun 2001 20:31:07 -0700

On Wed, Jun 13, 2001 at 07:23:00PM -0700, Marshall T. Rose wrote:
> hi. i don't think it makes sense to talk about a BEEP URL, because BEEP, by
> itself, isn't a protocol like HTTP or FTP or SMTP is.

Yes, though to be a little more accurate, BEEP isn't a service like HTTP
or FTP servers.  The URI scheme doesn't refer to the protocol.

> i suspect there is some general "issue" in the URL architecture that needs
> to get resolved before we will find a satisfactory answer to the question.
> for example, in the case of wsdl or soap or ipp or whatever, i can always
> say
>         http://example.com/
> but how is anyone to know that the http server there is really a front-end
> to wsdl or soap or ipp? my point is that there's something semantic going on
> here that isn't visible in the URL, instead you somehow have to know from
> context that if you see a URL inside a certain kind of message then there's
> an expectation that a special kind of process is sitting at the http server.

That is already true.  The "http" URI defines a namespace with hierarchical
characteristics that is resolved by communicating with the naming authority
(server) using one of the HTTP family of protocols, but such resolution
only takes place when it is used within specific references as a hypertext
link or submit button.  Otherwise, it is just an identifier.

They could define a beep URL that had the property of discovering which
profile to be used as part of the connection, but that isn't very useful.
URI schemes are more like TCP reserved ports -- they correspond to the
type of service rather than the specific protocol to use (i.e., port 80
is reserved for WWW, not HTTP/1.0).

> i noted that the wcip draft says this about naming:
>     wcip://" domain-name ":" port "/" channel-name "?proto=beep"
> but you could just as easily have "?proto=http" tacked onto the end.
> i'm not suggesting that this is the "right way" but it does bear some
> thinking.

YUCK!  wcip needs to learn about hierarchy violations.


would be better, but still misses the point.  The URI scheme should
answer the question of "what application interface should I expect?",
which is a a lot more than "what protocol should I use?".


Roy T. Fielding, Chief Scientist, eBuilt, Inc.
                 2652 McGaw Avenue
                 Irvine, CA 92614-5840  fax:+1.949.609.0001
                 (fielding@ebuilt.com)  <http://www.eBuilt.com>

                 Chairman, The Apache Software Foundation
                 (fielding@apache.org)  <http://www.apache.org/>