W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2010

Re: I-D Action:draft-loreto-http-timeout-00.txt

From: Salvatore Loreto <salvatore.loreto@ericsson.com>
Date: Fri, 09 Jul 2010 13:34:37 +0300
Message-ID: <4C36FB3D.9010109@ericsson.com>
To: Mark Nottingham <mnot@mnot.net>
CC: "Thomson, Martin" <Martin.Thomson@andrew.com>, Greg Wilkins <gregw@webtide.com>, HTTP Working Group <ietf-http-wg@w3.org>
On 7/9/10 10:43 AM, Mark Nottingham wrote:
> On 09/07/2010, at 3:50 PM, Thomson, Martin wrote:
>
>    
>> Right - for the cases you have in mind, the single bit might be wasted bytes too.
>>      
> Not really. Patching Squid (etc.) to change policy based upon the presence of a header isn't a big deal, and when it's deployed as an accelerator, the single bit could be quite valuable.
>    
if I understood correctly Mark, you are to suggesting that just having 
an header would be enough and much easier to implement for the 
intermediaries, isn't it?
if so I don't see much difference in defining an header containing a 
value and let the intermediate decide if consider or not the value and 
then use it as a single bit!

/Sal
>    
>> However, in the context of XmlHTTPRequest, might we not consider the browser to be an intermediary?  Would the same constraints apply there?
>>      
> Are you suggesting that the browser rewrite headers before handing them to the HTTP API? That's likely to cause a fair amount of confusion... Why not just expose the connection timeout of the browser as an API extension?
>
>
>
> --
> Mark Nottingham     http://www.mnot.net/
>
>    


-- 
Salvatore Loreto
www.sloreto.com
Received on Friday, 9 July 2010 10:35:37 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:23 GMT