> Magnus Ahltorp wrote:
> My request is to enable the IPv6 API whenever the IPv4 code is enabled.
...
> Because it is much simpler and cleaner to choose the implementation of
> the comatibility layer at compile time. Having to make this choice at
> runtime makes code inefficient and hard to maintain.
This would be reasonable if some IPv6 API really existed. 8)
Those one which exist now de-facto are not standard de-jure and vice versa.
So that, it is API that is experimental. Our IPv6 stack itself
is not more experimental, than IPv4 f.e. in. Linux-2.0.
Writing IPv4 apps. using IPv6 API you will get not-working apps, which
will loose binary compatibility as soon as some new bright idea will get
into head of some member of IPng WG. It is practice.
And theory is that correct application must be protocol independent
f.e. if it uses SOCK_STREAM, it must not depend, what kind of network
protocol was handed you by name server, protocol switch or another
host policy engine. It must work equally over TCP/IP, TCP/IPv6,
ISO stack or some another connection-oriented protocol. So that
even the question does not make much of sense, you need not IPv4
mapped addresses at all.
Alexey Kuznetsov
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/