The problem with the host-name-address parameter IMO is:
a) It servers little purpose since you can get through a
NAT like the nat draft says by just NOT placing any
addresses in your INIT.. yes you are singly homed.. but
the HNA param would not help.. you have to have nat's
cooperating to get M-homed..
b) It is very very difficult to implement in the kernel since
when it is time to resolve the host name one must
1) upcall to an app to do the DNS lookup
2) sleep the kernel thread waiting for an app
to write back
3) when the app gets the DNS response have it
write it down the "socket" or other special
IPC mechanism
4) Then proceed.
This is very ugly and so far, we have not been able to
test this feature at any interop since no one is implementing
it.. and for good reason... I really think it should be something
we drop from the protocol :-D
R
>>
>> Thank you in advance,
>>
>> chiba
>>
>>
>
> Forwarding to my cisco account so I can answer
>
> R
>
>
-- Randall Stewart ITD 803-345-0369 <or> 815-342-5222Received on Fri Jan 28 11:52:40 2005
This archive was generated by hypermail 2.1.8 : Mon Mar 13 2006 - 15:22:23 EST