Re: [IPFIX] Endianness of address types: 5101bis section 6.1.2.
Brian Trammell <[email protected]>
| Newsgroups | gmane.ietf.ipfix |
|---|---|
| Message-ID | <[email protected]> |
Hi, Paul, Hm. The lack of documentation on network byte order is odd, though I suppose the words "network byte order" are on their own sufficient for any reasonably competent implementor to know what we're talking about. So I think the right thing to do is be very clear that everything in IPFIX is network byte order (as you say, in section 3). Will go with Andrew's language, and review 5101bis to ensure that we're as clear about byte order as we need to be elsewhere. Many thanks, cheers, Brian On Oct 22, 2012, at 7:44 PM, Paul Aitken wrote: > Brian, > > +1 to what Andrew said. > > Additionally, please cite an xref for "network byte order". > > (Why isn't there an IETF terminology doc for these things?) > > Else, > > - RFC 5101 section 3? > - RFC 1886 or RFC4122? > - draft-newman-network-byte-order from '99 ? > > Thanks, > P. > > > On 22/10/12 18:39, Andrew Feren wrote: >> Hi Brian, >> >> I can't say found the original confusing, but if some people did I don't have a problem with the new form. Or even more simply. >> >> Address types -- ipv4Address, macAddress, and ipv6Address -- MUST be >> encoded as four, six, and sixteen octets in network byte order, respectively. >> >> -Andrew >> >> On 10/22/2012 07:08 AM, Brian Trammell wrote: >>> Greetings, all, >>> >>> I've received a comment off-list from an implementor that section 6.1.2 of 5101 (and 5101-bis) is potentially confusingly worded. The intent here is that all addresses be _encoded_ in big-endian order, regardless of how they are handled internally (e.g., Unix programs often use both host and network byte order for IPv4 addresses: the former for manipulation, the latter for compatibility with the sockets API). "6-octet integers" and "16-octet integers", however, are kind of weird concepts. So I'd suggest the following change: >>> >>> OLD: >>> >>> 6.1.2. Address Types >>> >>> >>> Address types -- macAddress, ipv4Address, and ipv6Address -- MUST be >>> encoded the same way as the integral data types. The macAddress is >>> treated as a 6-octet integer, the ipv4Address as a 4-octet integer, >>> and the ipv6Address as a 16-octet integer. >>> >>> NEW: >>> >>> Address types -- ipv4Address, macAddress, and ipv6Address -- MUST be >>> encoded the same way as unsigned integral data types, as four, six, >>> and sixteen octets in network byte order, respectively. >>> >>> >>> Thoughts? >>> >>> Thanks, >>> >>> Brian >>> >>> _______________________________________________ >>> IPFIX mailing list >>> [email protected] >>> https://www.ietf.org/mailman/listinfo/ipfix >> >> _______________________________________________ >> IPFIX mailing list >> [email protected] >> https://www.ietf.org/mailman/listinfo/ipfix _______________________________________________ IPFIX mailing list [email protected] https://www.ietf.org/mailman/listinfo/ipfix