Re: [tsvwg] draft-ietf-udp-options issues from IETF 104

[Derek Fawcus <[e-mail]>]

On Thu, Jul 18, 2019 at 08:08:01AM -0700, C. M. Heard wrote:
> 
> I think Derek mis-stated what I had in mind., which was was something
> similar to what's in RFC 8200 to specify what happens if an options-aware
> implementation encounters an option that it does not recognize:
> 
>       0 - skip over this option and continue processing the packet
> 
>       1 - discard the packet.
> 
> This processing instruction would apply to an individual option. The options
> would not be coupled together.

Maybe I'm not understanding something, but I see that as a distinction without a difference.

Assume some packet is received having legacy payload of zero length, options
X, Y, Z, and non-legacy data of length Q.

If options X & Z are understood, but Y is not, and Y has this magic bit set to 1,
I'd expect the whole of the surplus to be discarded.

So depending upon how one implements this, the non-legacy receiver will either
receiver a UDP packet of zero length (as a legacy receiver would),
or simply not receive anything (i.e id discarded the whole packet, not just the surplus).

i.e. the only coupling / fate sharing here is the fact that one of the options
implies the whole of the surplus is to be thrown away.

DF