I agree completely.
Regards,
Stan
-----Original Message-----
From: aodvv2-discuss-bounce@xxxxxxxxxxxxx [mailto:aodvv2-discuss-
bounce@xxxxxxxxxxxxx] On Behalf Of John Dowdell
Sent: Tuesday, June 23, 2015 9:14 AM
To: AODVv2 Discuss
Subject: [aodvv2-discuss] Some comments on RFC5444 usage draft
Hi all
First let me say I am definitely in favour of having this document [1],
because
it is something I badly needed in my early days of co-authoring AODVv2,
trying to map AODVv2 messages directly to RFC5444 packets (which I now
know was a Bad Thing to do).
Having had a first read, I cannot find the bit I really wanted to see, and on
which subject we have had many discussions in the AODVv2 author team;
exactly how to interface routing protocol messages to a generic
RFC5444 parser, through a defined API. Some may consider this to be an
implementation-specific matter, but given that at least three of us (and I
hope I'm not speaking out of turn) have now fallen into almost the same trap
of trying to do the Bad Thing, I'd really like to see an API defined.
Having re-read section 4 of the draft, I can see that it sort-of defines an
API,
but I'd like to see more. I recall we have discussed how <routing protocol>
should be able to inject messages that then go on to impact RFC5444 packet
content, but that there are parts of the packet structure that are reserved
for 5444 parsers and should not be messed with.
I'd welcome comments from y'all, since you are closer to the coal face than I
am on this topic right now.
I do wonder than when we're done discussing, we could send a consolidated
review to the manet list?
Best regards
John
[1] draft-clausen-manet-rfc5444-usage-00