David Stuart wrote:
> Hi, I have a follow up question to the question below.
>
> Let's say I send a request, which results in a redirection with the
> following contacts:
>
> 1. [EMAIL PROTECTED];q=1.0
> 2. [EMAIL PROTECTED];q=2.0
> 3. [EMAIL PROTECTED];q=3.0
First, q-values must be in the range [0.0-1.0] so I am going to assume
all your q-values have been divided by 10.
> Now, while handling the first of the above contacts, I get a second
> redirect with the following contacts:
>
> 1. [EMAIL PROTECTED];q=2.5
>
>
> Am I supposed to handle this new redirect after 2 and 3 above, or in
> between 2 and 3?
According to 3261 you should merge the last one in according to q-value
among whatever candidates are left when you get it. (Your 2nd choice.)
So the relative order would depend on how many of the others you have
completed.
But this has been discussed and considered problematic. The problem is
that q-values are only ordinals, not cardinals, so mixing values from
different sources makes no sense. Also, the order then depends on
whether proxies recurse or redirect.
RFC3841 specifies a different order. It corresponds to your first
choice. And it doesn't have the other problems. I would recommend you
follow that.
Paul
>>----------------------------------------------------------------------
>>
>>Message: 1
>>Date: Fri, 06 Jan 2006 13:31:27 -0500
>>From: David Stuart <[EMAIL PROTECTED]>
>>Subject: [Sip-implementors] qvalue question
>>To: [email protected]
>>Message-ID: <[EMAIL PROTECTED]>
>>Content-Type: text/plain
>>
>>Hello,
>>
>>If (let's say, in a 302 response) there are multiple contacts, some of
>>which have the q-value set, and others which do not, how do I treat the
>>contacts with no q-value? Are they considered to be higher or lower
>>priority than those with the q-value?
>
>
>
> Dave
_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
