On May 19, 2013, at 14:52 , Fernando Gont <[email protected]> wrote: > On 05/19/2013 06:27 PM, james woodyatt wrote: >> >> That's not the sort of assumption I want to see go unstated in a >> Standards Track RFC. Better would be to avoid the ambiguity >> altogether. For example, OS X does not have a requirement that >> network interfaces have unique names. > > How do you ifconfig(8) them if the name is not unique?
The names used by ifconfig(8) are temporally unique, but not universally unique, i.e. it is not the case that 'net0' at time t=0.0 ns is the same as 'net0' at time t=0.1 ns. When they are different, the UUID of the network service associated with the 'net0' interface is different. >> There is a UUID that >> identifies network services, which are related to the network >> interfaces the system configuration service manages, but network >> *services* on OS X are not the same as network *interfaces*, and they >> are not mapped 1:1 to one another. > > I'm kind of lost here, sine I don't know the details behind OS X. That's why I'm trying to help. > Are you arguing tha the discussion of interfae names (or something else) > is incorrect? Are you arguing that there's something that should be > clarified? (if so, what, and how?) > > ANy help in this respect will be really welcome. Yes, I'm saying that interface names don't map well to the concept you're expecting them to indicate. You seem to think that a networking interface interface used with a particular network service will always have the same name every time it is put into service. That's not the case with OS X, which makes no such guarantee. Yes, many people labor under the misconception that 'en0' is always the Ethernet port, but we try to disabuse them of that idea every chance we get. The same thing goes for the interface index. The only thing that really identifies the interface uniquely is the UUID for the network service that you find in the System Configuration Dynamic Store, but even that may not be good choice for the Net_Iface value here, because services come and go as users make configuration changes and migrate their systems from one operating system release to the next. Look, the more I think about this issue, the clearer it becomes to me that you just need to change... >>> o it MUST be different for each network interface ...to... >>> o it MUST be different for each network interface simultaneously in use. That will leave the OS X system configuration service with sufficient latitude that it can choose a value for each network interface it configures to meet the goals of this specification. -- james woodyatt <[email protected]> core os networking -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
