Re: issue 325: When are Location's semantics triggered?, was: Protocols/APIs and redirects

On 2011-12-07 16:43, Cameron Heavon-Jones wrote:
> On 07/12/2011, at 2:08 PM, Julian Reschke wrote:
>
>> On 2011-12-07 14:45, Willy Tarreau wrote:
>>> ...
>>> I don't think we should say something like this. For instance a new 3xx
>>> might provide a Location header which would only be conditionnally used.
>>> Requiring UAs to process 3xx like 302 if they see a Location header seems
>>> problematic for the long term.
>>>
>>> Why not say that a Location header might be used in conjunction with new 3xx
>>> headers but UAs must not use it unless they understand this 3xx ?
>>> ...
>>
>> Well, the question is what UAs should do with a 3xx + Location. AFAIR, Safari follows the redirect, while others do not.
>>
>> Safari's behavior is "good" as it would allow to define 308 (as the permanent version of 307), and nothing needs to be changed in the browser.
>>
>> Best regards, Julian
>>
>
> these test results might be of interest:
>
> http://www.cmhjones.com/browser-http-test-matrix.html

...so Safari follows an unknown redirect only if the response doesn't 
have a payload.

Best regards, Julian

Received on Wednesday, 7 December 2011 15:51:26 UTC