[ih] UDP Length Field?

[Joseph Touch]

Hi, Noel (et al.),

> On Nov 29, 2020, at 4:46 AM, Noel Chiappa via Internet-history <internet-history at elists.isoc.org> wrote:
> 
> I suspect that, plus i) belt and suspenders robustness - if the UDP length is
> not <= the length as given by the IP header, there's an error;

TCP would have needed that too - esp. given it is intended to be more robust AND the checksum isn’t affected by zero-padding.

> and ii) Vint's
> point that one can pad to achieve even word length. (I wonder what TCP does
> for that, without a length field? I guess it doesn't.)

This makes more sense to me, as it would allow use of machine word size padding for UDP packets, but assume that TCP did its own data-padding (sending machine wordsize segments) for all except open/close. I.e., it would reduce UDP processing overhead by allowing machine size transfers all the time - that seems very useful when doing realtime (as was its initial motivation).

Joe