Hi Carsten: > >Conceptually, decompression with the C bit set means that the elided >field is reconstituted, i.e. a UDP checksum is computed. >No difference between node and router.
[Pascal] Agreed >*As an optimization*, a node that is going to discard it right away >(a) when re-compressing on the other side of the forwarder or (b) when >checking the checksum, might not do the computation (I would expect >that optimization to be implemented everywhere, but it is not a >feature of the protocol). > [Pascal] Agreed. Also when passing the packet to ULP in clear understanding that a MIC takes place. >I believe the places where I used MUST should be MUST. They are >essential features to make the compression mechanism work. >When you say SHOULD, what special conditions are you envisioning here >where that would not be heeded but the protocol still works? [Pascal] Makes sense. Can you please copy/paste/update the proposal to restores MUST where you reckon it's mandatory? >Field bus protocols don't encompass the pseudo-header in their >checksum, so technically they are outside the condition that allows >the application (yes, I should have said upper layer) to indicate >authorization to elide. If they have their own addressing, that is >rarely out of a global address space like the IPv6 address protected >by the pseudoheader. [Pascal] A protocol should protect itself. If the field protocol is already IP/UDP then it does. For the tons of protocols in place in buildings and industrial spaces, I guess that the gateway that performs the tunneling will have to be configured to tell whether the tunneled protocol is intrinsically safe. Since we will not provide that list, the goal is to have a text that's just open enough to allow it. >If we turn some of the MUSTs into SHOULDs, we need to add text that >explains the purpose of the UDP checksum and the pseudoheader so >readers not as familiar with IPv6 get enough information to make an >informed decision whether they really should be violating the SHOULD. [Pascal] The ball is in your camp :) please propose a revision with MUSTs. Pascal _______________________________________________ 6lowpan mailing list [email protected] https://www.ietf.org/mailman/listinfo/6lowpan
