On Jun 10 2008, at 16:13, Pascal Thubert (pthubert) wrote:

> we still call the ISA100.11a transport
> UDP though it's not

OK, so this transport protocol would like to use an 8-byte header with  
a source port number, a destination port number, and two highly  
compressible fields (redundant length, and the now redundant  
checksum), but it has its own *end-to-end* checksum, which is better  
than 16-bit one's complement and *does* include a pseudo header.

Two observations:

1) compression works best when it is done cross-layer.  So one would  
want to compress the whole stack from IP upwards to the ISA100  
transport.  This of course could include knowledge about the layer  
above UDP, just as we have included knowledge about RTP in RFC  
3095/5225.

2) Given this transport, the end system is likely to want back the  
feature of UDPv4 that allowed sending a zero checksum.  This transport  
is a nice example for how it was a mistake to conflate policy and  
mechanism in UDPv6: Encoding the *policy* not to send packets without  
a checksum that includes a pseudo-header with the *mechanism* to never  
allow zero checksums in UDPv6.
Of course, you should not have mechanism that a policy makes sure you  
will never use, but the mechanism is already there for IPv4, and the  
example is exactly one such case where the policy would not imply non- 
use of the mechanism.

Obviously, if an end system does not want protection from UDP (and it  
shouldn't want it in the ISA100 case), it should say so by setting the  
checksum to zero.  The mechanism/policy conflation mentioned above  
makes that impossible.

Changing UDPv6 at this stage is a big step.  If that is not what we  
(the IETF/the IPv6 implementers community) want to do, there is no way  
for the end system to make that statement.  This does not mean that we  
shouldn't honor it.

Gruesse, Carsten

_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan

Reply via email to