Re: [tsvwg] Alternative version of the UDP FRAG option

[Derek Fawcus <[e-mail]>]

On Tue, Mar 12, 2019 at 06:08:51AM -0400, Joe Touch wrote:
> That is lite alone only AFAICT. And we already know that is the risk CURRENTLY. 
> 
> Again, we need to include fixing these errors. If we let every bug persist, nothing would be deployable anymore. 

The way I view it is that we need to get a useful sub-set of the UDP-Options
deployable, and that would break the chicken-vs-egg situation which prevents
getting other systems fixed.

So if say LITE alone requires that systems be fixed (which I think it will
due to not wanting to checksum the LITE data, yet needing the packet to be
checksum neutral), while sending segmented large UDP packets (the current
LITE+FRAG case) can be made to work in the face of the broken systems,
we shoukd do so.

I think that addition, namely restoring the ability to send large (> pMTU)
UDP payloads, which has been somewha broken by NATing of IP fragmented
UDP is worthwhile, and probably sufficient to bootstrap UDP-Options.
It is also an improvment on the IP fragmented case in that it each FRAG
has port information.

So we can probably make every use except the pure LITE data case deployable,
making them deployable should eventually result in systems being fixed,
such that pure LITE is usable.

DF