On Apr 22, 2010, at 3:50 AM, <L.Wood@xxxxxxxxxxxx>
<L.Wood@xxxxxxxxxxxx> wrote:
http://tools.ietf.org/html/draft-ietf-dccp-udpencap-00#section-3.3
section 3.3 of the draft and its use of UDP as UDP-lite -- despite
UDP-lite being given a different protocol number because consensus
was that messing with UDP checksum and length fields in this way was
not compatible with established UDP implementation use - makes me
very very uneasy. (The earlier phelan draft had a hack around using
a UDP zero checksum, which is arguably worse.) This draft cannot go
forwards. Or it should use UDP-lite. This is a workgroup draft?
Seriously? Section 5 of RFC3828 and tsvwg experience indicates why
this is a bad idea. Is expedience more important?
Some others have made this comment, too, so I think this is something
that should be changed before the draft moves forward.
None of the drafts in question are final yet, so I think the
discussion and the subsequent decision should be based on the merits/
disadvantages of the overall approaches, rather than the current state
of the drafts. Although I'm sure the authors will appreciate all
detailed comments on their drafts as well.
- Pasi
