Re: [nat66] Comments on draft-mrw-nat66-12

[Fred Baker <[e-mail]>]

On Mar 15, 2011, at 10:28 AM, Rémi Després wrote:

> 3.2 and remainder of the document.
> The word datagram seems to be used instead of packet:
> - RFC 2460 doesn't use the word datagram for IPv6, even in case of fragmentation 
> - In any case, NPTv6 operates individually on packets without concern with reassembling fragments. 

You're correct that I used "datagram" interchangeably with "packet". Are you asking me to explicitly globally replace "packet" with "datagram" for consistency? I can do that.

This is a point where people often get into hissy fits and I start seeing double as my eyes cross and roll to the back of my head. "Packet" is a general term for "a bit of data wandering around as a quantum", but in X.25 refers to a PDU at the packet layer. Sometimes it's called a "frame", usually at the link layer. TCP calls it a "segment", DCCP calls it a "packet", SCTP sometimes calls it a packet and sometimes refers to "payload", and UDP calls it a "datagram". There are various words used by various applications. Anything ISO calls it a Protocol Data Unit or PDU, and may prepend a word/letter to indicate the layer in question; TP4, for example, exchanges TPDUs. But of course, since many PDUs in fact are parameters of API events as opposed to messages in the network, I tend to think of a PDU as a quantum of data passed through an API.

There are 1215 RFCs that refer to a packet containing an IP header and whatever happens to be inside it as a "datagram"; there are 2814 RFCs that refer to "packets", many of them in an IP context. RFC 1594 defines "datagram" as "a self-contained, independent entity of data carrying sufficient information to be routed from the source to the destination computer without reliance on earlier exchanges between this source and destination computer and the transporting network." http://searchnetworking.techtarget.com/definition/datagram has a fairly clean definition of the term. I find myself referring to IP's messages specifically as "datagrams", and quanta of data at any layer as "packets" pretty generically, and in any Internet context to view the two terms as largely interchangeable - as do, it would appear, 6028 of my colleagues that write RFCs. And no matter what I call them, I get a comment from someone telling me they wished I would call them something else.

I'll do the global replace.