Re: Evaluation: draft-ietf-tsvwg-udp-lite - The UDP Lite Protocol to Proposed Standard

[Erik Nordmark <Erik.Nordmark@sun.com>]

                    Yes    No-Objection  Discuss *  Abstain  
Erik Nordmark       [ X ]     [   ]       [   ]      [   ] 

First a question not directly related to the document.
It would be useful if *n*x implementations had the same IPPPROTO_foo
name for UDP-Lite. Typically this is derived from the keyword
in http://www.iana.org/assignments/protocol-numbers
But having they keyword be "UDP-lite" means there needs to
be a mapping since "-" isn't allowed in the identifier.
Is anybody concerned about this?
Would it make sense to request in the IANA section that the keyword
be e.g. "UDP_LITE" or "UDPLITE" to increase the probability that
a particular "foo" gets chosen to increase portability?
Or will there be a separate API informational document for UDP Lite?


Nits:
Section 3.1 second paragraph: It would be useful to point out that independent
of the coverage value the pseudo-header is always included in the checksum.

Section 3.1 third paragraph: It makes sense to add the standard 
      For computing the checksum, the checksum
      field should be zero.

Section 5 second paragraph: The text is unnessarily IPv4 centric.
Instead of saying "ICMP Protocol Unreachable" how about
"ICMPv4 Protocol Unreachable or ICMPv6 Payload Type Unknown (based
on IP version)".
Same parargraph: I think 4.4BSD code did not generate a protocol unreachable
but just silently dropped the packet. I haven't checked if this is still
the behavior in BSD code.
Has the WG made that check?

   Erik