Re: Using fragment identifiers with URNs

On Thu, Sep 27, 2001 at 07:44:12PM -0400, Al Gilman wrote:
> At 06:13 PM 2001-09-27 , Roy T. Fielding wrote:
> >
> >To correct an earlier statement, both the mailto and telnet schemes, when
> >used in Web browsers, are operated upon using GET semantics -- the resulting
> >retrieval is a form for composing a mail message or an application with an
> >open session.
> 
> You may wish to look on it that way.  But that is an implementation detail in
> the coding of the User Agent.  The intrinsic transaction that goes with the
> mailto: URI is the PUT to the mailbox.  The rest is at the election of the
> implementer.

Nope.  When you select a link on the Web it results in a GET action.
When the user finishes composition, they perform a POST action.  Those
are two different actions using the same identifier information.  This
dates waaay back to the 1992 architecture discussions.  That's how the
semantics were defined to be consistent with the rest of the Web and to
prevent unsafe posts from masquerading as links.

> Get Me My Mailtool is inserted in the transaction stream by the User Agent
> shell you call the browser.  It is not implied by the URI itself.  The URI is
> not 'retrieved' until the composition session is done and the message is in
> the
> hands of a competent MTA.

I don't know where you get that idea.  libwww and libwww-perl implemented
it exactly as I said.

....Roy

Received on Thursday, 27 September 2001 20:01:11 UTC