Wietse Venema wrote the following on 12/19/04 23:39 :
>>In short, do I need to implement a full-fledge IPv6 address parser when
>>I try to manipulate IPv6 addresses or is the format used a subset more
>>usable by a computer ?
>>
>>
>
>No official Postfix release has IPv6 support.
>
>The third-party code that I am building into Postfix, and that will
>hopefully ship with Postfix 2.2 when it becomes the official release,
>uses the inet_ntop() routine.
>
>
Thanks for the details, it could very well help me debug some odd
problems in the near future :-).
>Since I have no plans to bypass system library routines, Postfix's
>result of address to string conversion will be whatever the local
>inet_ntop() implementation produces.
>
Makes sense.
Policy daemons can then call inet_pton and handle the mess from this point.
> This is a member of a relatively
>new group of functions, and the manual pages do not say much about
>the exact output format of IPv4 in IPv6 addresses.
>
>
Hum, pleasant situation, guess I'll have to mark some of the greylisting
algorithms as experimental with IPv6 in SQLgrey for a while...
What did the people behind IPv6 thought when they described address
representations, nobody told them that DNS was designed just to solve
the human-representation problems they didn't solve ?
In the end it makes more sense to deal directly with the original binary
representations.
Lionel.
|
