Re: [tsvwg] Comment on "draft-ietf-tsvwg-udp-options-28" -> do we need to update the text oin the Length Field

[Gorry Fairhurst <[e-mail]>]

On 22/12/2023 17:55, Joe Touch wrote:
> No need to- it’s actually correct.  A udp payload length of zero is a 
> udp length of 8.
>
> I’ll put that in as a clarification via this thread.
I think it would be helpful to change this sentence to clarify things;

OLD:

"The UDP option area can be used with any
UDP payload length (including zero)"

NEW

"The UDP option area can be used with any length of UDP user data
(including zero).
A datagram with zero-length user data results in a UDP Length field
value of 8."

- The cost of being clearer is little for the draft and avoids this 
resurfacing.

Gorry


>
>> On Dec 22, 2023, at 9:05 AM, C. M. Heard <[e-mail]> wrote:
>>
>> 
>> On Fri, Dec 22, 2023 at 8:51 AM Gorry Fairhurst wrote:
>>
>>     On 22/12/2023 15:18, C. M. Heard wrote:
>>>     On Fri, Dec 22, 2023 at 1:45 AM Gorry Fairhurst wrote:
>>>
>>>         ... <snip discussion on options for jumbograms> ...
>>>
>>>         I have two questions, intended to help focus on any needed
>>>         changes:
>>>
>>>
>>>         1) I found one instance of text in the current draft that
>>>         might benefit
>>>         from a change, to avoid a potential mis-read:
>>>
>>>         The current draft contains: "The UDP option area can be used
>>>         with any
>>>         UDP payload length (including zero)",
>>>
>>>         - If I understand, I think it would be helpful to clarify
>>>         that this is
>>>         the case of "zero-length user data", rather than the length
>>>         field of the
>>>         UDP header, e.g.:
>>>
>>>         "The UDP option area can be used with any length of UDP user
>>>         data
>>>         (including zero).
>>>         A datagram with zero-length user data results in a UDP
>>>         Length field
>>>         value of 8."
>>>
>>>
>>>     That sounds right to me.
>>>
>>>
>>>         2) Does this draft also need to deal with UDP Length = 0?
>>>
>>>
>>>     No.
>>>
>>>
>>>         I understand, the length field is zero, as a special case
>>>         for the IPv6
>>>         Jumbogram, but I suggest this could be a seperable topic.
>>>         One suitable
>>>         approach could be to keep the updates to define this with
>>>         the PS spec
>>>         that defines the change in the UDP processing, and for
>>>         example to only
>>>         note that:
>>>
>>>         "If UDP options is defined for UDP datagrams with a length
>>>         field less
>>>         than 8, a specification would need to to define whether and
>>>         how this
>>>         format identifies the presence of UDP options."
>>>
>>>
>>>     I don't think we should do even that. The proposed text in
>>>     https://github.com/tsvwg/draft-ietf-tsvwg-udp-options/issues/26
>>>     adequately addresses the issue raised by IPv6 Jumbograms,
>>>     which is that in that case the UDP packet has no independent
>>>     UDP Length, and the fact that this arises from UDP Length == 0
>>>     is incidental.
>>>
>>>     Given there are no other known or proposed uses of values
>>>     of UDP Length <  8, I think we can leave it at that.
>>>
>>>     Mike Heard
>>>
>>     I'd personally be happy with both of your suggerstions!
>>
>> Perhaps, then, open a github issue for #1 above?
>>
>> Mike