Hi Bill,

On 2012-03-07 01:08, Bill Fenner wrote:
> On Mon, Mar 5, 2012 at 7:15 PM, Brian E Carpenter
> <[email protected]> wrote:
>> Carsten,
>>
>> On 2012-03-06 12:22, Carsten Bormann wrote:
>>> On Mar 6, 2012, at 00:00, Brian E Carpenter wrote:
>>>
>>>> No, I think it's exactly *not* confused on this point. There's
>>>> a distinction between the idealised URI and the produced URI;
>>>> in the produced URI, "%25" stands for "%" in the idealised URI.
>>> Ah, so the ABNF is wrong
>> I don't believe so. The ABNF does not describe the produced (encoded)
>> URI. I have read sections 2.2 and 2.4 of RFC 3986 several times
>> before asserting this. Of course I could be wrong, but we are waiting
>> for a review from [email protected] who will hopefully give a
>> definitive answer.
>>
>>> and the (vague) text is meant as I read it first (it can be read in other 
>>> ways).
>> Please indicate exactly which part of the text is ambiguous, and we'll
>> change it.
>>
>>> Replace
>>>
>>>       IPv6addrz = IPv6address [ "%" ZoneID ]
>>>
>>> by
>>>
>>>       IPv6addrz = IPv6address [ "%25" ZoneID ]
>> No. In the encoded URI that would end up as %2525. That's exactly the
>> trap that RFC 3986 warns against.
>>
>>>
>>>> We have no real choice but to use % since that was chosen years
>>>> ago, and that means that the produced URI contains %25.
>>> I don't know that.  You could use "percent" and that would work, too.
>> We could use any unreserved symbol, but that would need translation from
>> the RFC 4007 format. For example we could just use Z, as in
>> http://[fe80::aZen1].
>> Or we could use ~: http://[fe80::a~en1].
>>
>> Would that be less confusing?
> 
> Previous joint work between the ipv6 working group and the W3C URI
> working group resulted in a decision that did not change the ABNF at
> all, in 2 ways:
> 
> 1. It used the IPvFuture extension mechanism;

Hmm. I'm not sure that was a good choice. It certainly doesn't
seem natural to me. Also it seems like more work for implementors
than a small extension to the existing ABNF. What's so sacred?

> 2. It used a non-percent character for the separator.

Yes, and I have almost convinced myself that is better. The % is
certainly the worst possible choice from a clarity viewpoint.

> At the time, the URI working group was very concerned about the change
> in the ABNF and the use of percent where percent had not previously
> been allowed.  Have they changed their position here, or have they not
> had a chance to comment on this change yet?

It's been sent for URI review but no reply so far.

> 
> This work was accepted as an ipv6 wg work item around IETF63 (Paris,
> 2005), but the authors never pushed the document forward due to a lack
> of interest in the broader community.  The draft that was adopted by
> the wg was
> 
> http://tools.ietf.org/html/draft-fenner-literal-zone-01

Recently there's been definite interest in this as an operational
convenience. I was pretty much unaware of the 2005 work, since that was
just when I joined the IESG and life was too busy, especially
during the first Paris IETF.

Was there a real reason that you went for this?
  IPv6zone-id = 1*( unreserved / sub-delims / ":" )

We have
  ZoneID = 1*( unreserved / pct-encoded )
which allows anything, having noted that some operating systems use all
sorts of characters in interface names.

Thanks

     Brian
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to