On 2012-02-08 01:58, Juergen Schoenwaelder wrote:
> On Tue, Feb 07, 2012 at 11:05:20AM +0100, t.petch wrote:
>> Juergen
>>
>> This topic kicked off with
>> http://www.ietf.org/mail-archive/web/ipv6/current/msg14975.html
>> as  a report of some unexpected behaviour in Firefox and the view
>> there was that BSD and Linux imposed a limit of 15 characters.
> 
> While 15 characters seem to be a common length restriction in todays
> Unix implementations, there does not seem to be an architectural
> constraint that it has to be 15 - the socket API functions say there
> is a limit but they leave it open what it is. On a really low-end
> Juniper router in our lab (without doing anything to it), I find
> interface names like sp-0/0/0.16383 - already 14 characters. But then
> I also realize that this interface name contains characters that do
> not fit the unreserved production of RFC 3986 either.
> 
> So in short, I think we should avoid putting up length restrictions
> and we may need to think what to do about interface names that contain
> characters not matching the unreserved production of RFC 3986. 

I don't see any particular value in limiting the ZoneID string to 15.

Reserved characters can always be % escaped; that's not an issue,
but should be mentioned in the text.

> Perhaps
> if all characters are digits, we can treat the value as an interface
> number as a last resort.

The interface number is uint32 anyway, in the MIB and the socket API,
which limits it to decimal 4294967295, i.e. ten characters. In any
case the name/number mapping is host-specific and has nothing whatever
to do with the URI format.

   Brian


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

Reply via email to