Re: Concurrent non-error response disallowed. ("clarification of 7.2.2. Monitoring Connections for Error Status Messages")

On Sun, May 16, 2010 at 1:48 PM, Henrik Nordstrom
<henrik@henriknordstrom.net> wrote:
> lör 2010-05-15 klockan 03:13 -0700 skrev Wenbo Zhu:
>> So at least we agree that there are valid use cases for supporting
>> concurrent responses; and the spec at least hasn't explicitly
>> disallowed such a behavior.
>
> Spec allows early responses, and in specific requires clients to act in
> specific manner if the early response is an error response.
>
> the fuzzy spot is if servers can expect the rest of the request to be
> sent to them even if they finish the complete response before receiving
> the complete request.
>
> then there is a number of implementation details like store & forward
> operating proxies etc which will have impact on the delivered timing, so
> you cant rely on the early aspect to actually be delivered. It may well
> be that the server won't receive anyting (not even a connection) before
> the HTTP request is completely sent by the client.
A more general question is how well proxies support chunk-encoded
requests today, since browsers don't generate chunk-encoded requests.
Would like to hear your experiences on this

Thanks,
Wenbo



>
> Regards
> Henrik
>
>

Received on Monday, 17 May 2010 00:31:16 UTC