Re: [6lo] Intdir last call review of draft-ietf-6lo-minimal-fragment-04

[Dave Thaler <[e-mail]>]

From: Pascal Thubert (pthubert) <[e-mail]>
Sent: Tuesday, November 12, 2019 7:50 PM
To: Dave Thaler <[e-mail]>
Cc: [e-mail]; [e-mail]
Subject: Re: Intdir last call review of draft-ietf-6lo-minimal-fragment-04
Le 13 nov. 2019 à 09:02, Dave Thaler <[e-mail]<mailto:[e-mail]>> a écrit :

Pascal wrote:
[…]

   of the destination.  Upon a first fragment, a node creates a state
   and forwards the fragment.

I’d recommend “a node *attempts to* create state and forward the fragment”,
since either of those could potentially fail, right?

Also should this use normative language (“a node MUST…”)?

The state is then used to forward the
   next fragments of the datagram.



Ø  I’d suggest MAY attempt to create the state and if successful MUST use it to forward the next fragments.

To me, that sounds like an inappropriate use of MAY.  MAY means an implementer can choose to ignore it.
Do you really mean that an implementer can choose NOT to attempt to create the state?
If it doesn’t implement attempting to create the state, then how could the mechanism work at all?


I intended the MAY to indicate that the FF mechanism is optional. The node can always do what it did so far and either recompose the full packet to forward it or drop the packet for reasons of its own.

It could split in 2 sentences; say that the mechanism is OPTIONAL and then that when it is used the node MUST attempt to create the state; would that clarify ?

Yes.  Of course your entire document is optional, so I’m not sure you need to explicitly say it’s OPTIONAL as long as you don’t Obsolete any other RFC.
You could just say that “An implementation that does fragment forwarding MUST attempt to create the state…”