RE: [dhcwg] questions on RFC 3046 and RFC 3527

> Thanks for your response.
> 
> > Because the RemoteID isn't necessarily an IP address? 
> > At least one well-known standard (DOCSIS) sepcifies that 
> the Remote ID is
> the MAC address of the cable 
> > modem, and not an IP address.
> 
> It is true that RemoteID is not necessarily an IP address, 
> but it MAY be an
> IP address. The question is, if the RemoteID is an IP 
> address, is the use of
> RFC 3527 redundant or there are some other reason to include the Link
> selection sub-option? 

However, the DHCP Service cannot assume that the RemoteID is an IP Address.
All it knows is that there's this binary blob of data to be used for certain
logic, that that the RemoteID is unique (See section 3.2 of 3046).  And, how
would the service distinguish between an IP address and any other 4 byte
Remote ID?

> Regarding the confusion between RFC 3046 and RFC 3527 (Q2), I 
> think RFC 3527
> should say: 
> 
> "If the DHCP server supports the Relay Agent Information 
> Option the DHCP
> server MUST allocate the address based on the following when 
> the received
> Relay Agent Information Option contains the Link selection sub-option:
> "

I don't understand the confusion.  All 3527 says is that IF a DHCP server
claims to implement 3527, then the server must perform the address
allocation in the specified manner.  And since 3527 cannot be implemented
without implementing 3046, having 3046 support is implied.