On 2010-01-07 13:21, Juergen Schoenwaelder wrote:
> On Wed, Jan 06, 2010 at 05:35:21PM +0100, Dan Wing wrote:
>
>>> My main interest is textual comparability of addresses coming from
>>> potentially many different sources. If I can compare addresses safely
>>> only if I know some context information communicated out of band or I
>>> have to configure all sources to produce the same, I feel somewhat
>>> uneasy. My preference thus is 1), I might be OK with 2) if the
>>> frequency of the introduction of new WKPs is very small.
>> BEHAVE is introducing both (1) a WKP which embeds the IPv4 address
>> in the last 32 bits and (2) allowing networks to use their own 96-bit
>> prefix and embed the IPv4 address in the last 32 bits.
>>
>> Your proposal would prohibit (2) from displaying the IPv4 address in
>> dotted decimal, because of a fear the parsed IPv6 address might be
>> seen or used on another network.
>>
>> Personally, I find this useful. If I'm on my home network and somebody
>> sends me a trace with their IPv6 prefix and some dotted-decimal displayed
>> in the last 32 bits, I know immediately that their tool (running on their
>> network) believed -- through whatever means -- that the last 32 bits
>> was an IPv4 address.
>
> My concern are tools/applications generating different textual
> representations of the same addresses because of different contextual
> information they have. This then requires smart tools that can
> normalize things in order to make comparisons easy. Avoiding the need
> of special tools to enable comparisons of IPv6 addresses I think was
> one of the motivations behind this work.
Surely we can never hope to eliminate the need for converting to a
a canonical form before comparing. The canonical form for machine
comparisons is fairly obvious - a 128 bit binary number. The question
seems to be whether we can define a canonical form for eyeball comparisons.
I thought the idea of this work was to help humans by minimising the
variability in the "printed" version of an address. Showing an embedded
IPv4 address in dotted decimal is part of that, but it's only going to
happen when the source knows algorithmically that it *is* an IPv4 address.
I think we have to accept that this is not always the case, and that
there will inevitably be cases where two sources of a text representation
of the same address present different results, because one source has
contextual information that the other doesn't.
Brian
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
