Re: EDNS: OPT and forwarders

[Robert Elz <kre@MUNNARI.OZ.AU>]

    Date:        Wed, 01 Apr 1998 11:28:44 -0800
    From:        Paul A Vixie <paul@VIX.COM>

    suggestions which were raised during the meeting were:
[...]    
    suggestions i've thought of since then are:
    
    3. require that EDNS not be used with a forwarder unless the client's config
    data indicates that a given forwarder does in fact understand EDNS.

    i'd like to require "3" but suggest "2" at the implementor's discretion.

I'd prefer none of that.   Rather, I'd prefer a paragraph pointing out the
potential problem, and indicating that using eDNS in cases where it might
fail is unwise, and then lave it to the implementors to figure out what is
the best solution for them.

I would also not use the word "forwarder" anywhere in relation to this,
that has mental connotations in many people's minds, I would suggest simply
indicating that this can apply to any recursive query.   That is, if RD is
off, sending eDNS is safe (by design).  If RD is on, sending eDNS is only
safe if the resolver has reason to believe that server supports eDNS,
however that information is gained (config, an eDNS query with RD off, ...)

kre

ps: splitting longest match query into a separate draft is probably the
right idea, simply because by the time it gets some examples and such
it is likely to be several pages - but it does need to be written soon, so
Matt can use it for his IPv6 drafts, which the IPv6 community needs to
be at PS asap.   Paul, will you do that, or Matt, or do you want some other
volunteer?