Re: OPT == type 41?

["Donald E. Eastlake 3rd" <dee3@torque.pothole.com>]

I think it should be re-assigned.  I don't think the RFC can be
re-issued, although that's for the RFC editor to decide.  If not, a
new one should be issued obsoleting the current one.

I can update the IANA DNS considerations draft to indicate that some
implementations don't cache 128-255...

Donald

From:  Brian Wellington <bwelling@tislabs.com>
Date:  Tue, 7 Sep 1999 14:46:23 -0400 (EDT)
To:  namedroppers@internic.net
Message-ID:  <Pine.LNX.4.10.9909071444050.2818-100000@spiral.gw.tislabs.com>

>RFC2671 has assigned the OPT record type code 41.
>
>The OPT record is a meta/pseudo record, in that it cannot be cached.  
>Therefore, like other meta/pseudo records (TSIG, TKEY), it should be
>assigned a type code value in the 128-255 range.  This non-cacheability
>cannot be derived from the TTL field, since OPT overloads the TTL field to
>contain version, extended rcode, and flags.
>
>No document formally specifies 128-255 as the range for meta types.  BIND
>8 enforces this, and it seems to be current practice.  Should this be
>formalized?  It would be nice to have this specified somewhere.
>
>Should the OPT be renumbered (to 248, for example) and RFC2671 be
>reissued?
>
>This is an urgent issue.  There are no released implementations of EDNS0
>as of yet, but there will be soon.
>
>Thanks...
>
>Brian
>