Brian E Carpenter wrote: > There are multiple references to network byte order in RFC 3542, > and it's very specific for raw sockets. That seems clear enough. > > RFC 1958 says: > > 3.13 All specifications should use the same terminology > and notation, > and the same bit- and byte-order convention. > > (sadly, without saying what it is...) > > There really doesn't seem to be any scope for doubt here.
Thanks for the parenthetical levity, Brian. Unfortunately, 3.13 is somewhat confusing too, because by mentioning bit order, which is clearly different among link layers, it implies that the "convention" mentioned only applies to how bits and bytes are shown in a diagram. Not what the transmission sequence of bits or bytes ought to be. Clearly, for example, the MSbit is sent first in FDDI, last in Ethernet. And yet the MSbit is always shown on the left and numbered Bit 0 in Internet convention. But byte transmission order is a different matter. Typically, MSbytes are shown on the left *and* are transmitted first. (Not that I want to second-guess what you intended in that RFC.) I can certainly accept the idea that in 1981, when RFC 791 was written, there was no preference for big endian byte transmission order. Certainly not in DoD interface standards of the time, which I'm sure DARPA was well aware of. By now, it *should* be common knowledge. But I emphasize should for a reason. Thanks again. Bert -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
