[rfc-i] Relation to other RFCs - Updates

Martin Rex mrex at sap.com
Thu May 3 15:48:57 PDT 2012

Joe Touch wrote:
> On 5/3/2012 9:43 AM, Martin Rex wrote:
> ...
> > so for consistency,
> >    IPv6 does not "Obsolete" (=completely replace) IPv4,
> TCP-AO obsoletes TCP MD5. It is not backward compatible, but is intended 
> as its replacement.
> As others have noted, this is another use of "obsoletes", i.e., "use 
> this [complete] spec, not the obsoleted one".
> So we could have decided that IPv6 obsoleted IPv4 (even though we 
> didn't). The point isn't backward compatibility; it's whether the 
> issuance of the new protocol is intended as a replacement for the old 
> one AND the old one should *now* no longer be used.

Yes, there are two distinct usages of the term "obsoletes":

  (1) the natural language use and meanding of the word "obsoletes"
      in the description of the "Historic" classification
      in rfc2026 (IETF Standards process)

  (2) the narrowly scoped meaning of the "Obsoletes:"/"Obsoleted by:"
      as defined by the RFC Editor in rfc2223 and

And we should stop using (1) to mess up the navigational hints (2)
for the RFC readers.


More information about the rfc-interest mailing list