- From: Roy T. Fielding <fielding@gbiv.com>
- Date: Thu, 9 Sep 2004 13:41:39 -0700
- To: Mike Brown <mike@skew.org>
- Cc: uri@w3.org
On Sep 9, 2004, at 1:21 PM, Mike Brown wrote:
> I'd like to see the 2nd paragraph in the Introduction ("This document
> obsoletes [RFC2396]...") explicitly state whether the specs that
> reference RFC
> 2396 are expected to conform to RFC 2396bis in whole or in part, or
> whether
> such conformance is deferred until those specs themselves are revised.
"obsoletes" already says all that needs to be said. You are confusing
two entirely different things. There is only one standard here (URI).
The text you quote from 2026 is only applicable for something like a
new XRI standard that is intended to perform the same function using a
different protocol.
> I am somewhat confused about this issue, because previous discussions
> on the
> list led me to believe, for example, that the algorithm for resolving
> relative
> reference to absolute form would be applicable immediately, whereas
> more
> recently it was mentioned that RFC 2396's BNF productions would have
> to remain
> in effect as well, for the benefit of the standards that rely on them.
They are permanent documents! They hang around forever, even when
obsolete. The references to the old documents do not disappear
just because a new version comes out. The only thing that changes
is how new implementations are created and new protocols are defined
or updated.
> Given that RFC 2396bis's Introduction says it obsoletes RFC 2396, I
> assume the
> intent is to move RFC 2396 to Historic, and thus (according to RFC
> 2026 sec.
No, it is moving RFC 2396 to Obsolete status. That means implementors
are
requested to implement according to the new RFC regardless of how it may
or may not differ from prior RFCs. It is still the same standard.
Think of it as yards versus meters. Yards is a common form of
measurement
and was *the* standard for a very long time, including through several
iterations where the actual length of a yard was adjusted. Then the
meter came along as a replacement for *the* standard length. That was
a second standard along the lines that 2026 describes. rfc2396bis is
merely making changes to the documentation of the one URI standard
that we all know and use on a regular basis, and conforms better to
the existing implementations than RFC 2396.
....Roy
Received on Thursday, 9 September 2004 20:42:14 UTC