Re: mech-v2: unidirectional tunneling removed, SLLA/TLLA changes

[Francis Dupont <Francis.Dupont@enst-bretagne.fr>]

 In your previous mail you wrote:

   Two status issues on certain mech-v2 features:
   
   1) Based on Chirayu's comments earlier, I've changed the rule (from
   SHOULD NOT) to MUST NOT send TLLA/SLLA options on configured tunnels.  
   I believe some very old implementations, ages ago, did this, but it
   seems to make sense to get them fixed if they want to conform to the
   new specification. (See below.)  This does not create a severe interop
   problem as long as the other end implements the "MUST ignore" as well.
   
=> I object to this change because some usages were found (so could be found
in future) to IPv4 addresses in TLLA/SLLA: "SHOULD NOT" is enough.
An example is the (expired, ask if you can't find it) draft
"draft-ietf-ngtrans-hometun-01.txt" which was implemented and IMHO
has still interested waiting for the MOBIKE stuff which should provide
a better solution.

   If you have any objections, please raise them soon, and provide 
   suggested alternatives.
   
=> keep previous text and if you'd not like to mix SHOULD and MUST
use the more open one, i.e., SHOULD.

Thanks

Francis.Dupont@enst-bretagne.fr

PS: a MUST is incompatible with 6over4 too and we have not free time to
kill (i.e., make historic) 6over4 and all.