Re: [httpapi] client-specified timeout?

[Mark Nottingham <[e-mail]>]

Something like:

GET /foo HTTP/1.1
Host: example.org
Prefer: wait=2

?

<https://tools.ietf.org/html/rfc7240#section-4.3>

IIRC that was specified in seconds, not ms, because of concern that very small periods desired by the client would effectively be a race with the network; e.g., if your client wants a result in 200ms and you make a request to a server even mildly far away (due to connection setup times, etc.), the response is likely to take longer.

If we wanted something with finer granularity, we'd probably want to caution against such assumptions on the client side -- i.e., that 200ms is *server-side* processing, and the client shouldn't try to enforce the timeout. 

It might also be interesting to think about sending a 103 Early Hints response with a Preference-Applied header in some circumstances, so that the client knows the server is honouring the preference...

Cheers,


> On 6 Feb 2021, at 5:54 am, Salz, Rich <rsalz=[e-mail]> wrote:
> 
> Is it reasonable for a client to want the ability to say things like “if this takes more than 300ms send a failure status code back” ?
> Anyone in the WG interested in thinking/working on this?
>  
> -- 
> httpapi mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/httpapi

--
Mark Nottingham   https://www.mnot.net/