Re: #428 Accept-Language ordering for identical qvalues

[Julian Reschke <[e-mail]>]

On 2013-01-17 09:26, "Martin J. Dürst" wrote:
> Hello Julian,
>
> On 2013/01/17 0:56, Julian Reschke wrote:
>> Hi there,
>>
>> with <http://trac.tools.ietf.org/wg/httpbis/trac/changeset/2119#file1>,
>> the spec now says:
>>
>> "If no quality values are assigned or multiple language tags have been
>> assigned the same quality, the same-weighted languages are listed in
>> descending order of priority."
>>
>> This is a change from both RFC 2068 and RFC 2616 which we *did* discuss
>> back in the thread starting with <​
>> http://lists.w3.org/Archives/Public/ietf-http-wg/2011OctDec/0223.html>gt;;
>> back then we decided not to make this change because we know of
>> implementations ignoring the ordering, and no convincing argument was
>> given for making the ordering significant.
>
> We also know of implementations, both on the sender and on the receiver
> side (as far as I remember) that use ordering. So maybe something like:
>
> "If no quality values are assigned or multiple language tags have been
> assigned the same quality, the same-weighted languages may be listed in
> descending order of priority."
>
> I'm sure somebody can come up with better wording, but I hope you get
> the idea.

The key question is whether the order of same-weighted values matters. 
The proposed text doesn't answer that.

It hasn't mattered according in RFC 2068 and 2616, it doesn't matter for 
other Accept-* header fields, and we know of recipients ignoring the 
order. I think this is sufficient reason to stick with what we had in 
2616, meaning reverting this change.

Best regards, Julian