Re: EDNS standard clarification

[Michael Sawyer <Michael.Sawyer@nominum.com>]

On Tue, 10 Oct 2000, Bill Manning wrote:

> 	I would expect the behaviour to be "NOTIMPL" instead of a
> 	"silent" ignore. Its not a FORMERR, is it? :)

The problem I forsaw with something like that was what to do on getting
the NOTIMPL reply back, if you sent out a query with (for example) 10 OPT
attributes.  Unless you have something which indicates *which* of the 10
you didn't recognize, the client has no way of knowing which ones to turn
off, and turning all off just because one isn't implimented seems like a
bad idea.

Also, as Brian pointed out when we were discussing this the other day,
some of the OPTs may be just fine to ignore (key or encryption types we
understand, etc...).

I can think of arguments both ways, actually.  It may work to extend the
EDNS definition of the opt codes such that the upper bit means "you MUST
understand me or fail," or something along those lines.  I know that's
kind of ugly, but it may one way to deal with this.

					Mike




to unsubscribe send a message to namedroppers-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.