- From: James Snell <jasnell@gmail.com>
- Date: Mon, 5 Dec 2011 15:50:25 -0800
- To: Martin Thomson <martin.thomson@gmail.com>
- Cc: Julian Reschke <julian.reschke@gmx.de>, HTTP Working Group <ietf-http-wg@w3.org>
Oh.. the conditional request is definitely an interesting case. I'd definitely be open to adding some additional discussion. On Mon, Dec 5, 2011 at 3:46 PM, Martin Thomson <martin.thomson@gmail.com> wrote: > On 6 December 2011 07:29, Julian Reschke <julian.reschke@gmx.de> wrote: >> On 2011-12-05 21:20, James Snell wrote: >>> Ok... well like I said, I don't have a problem pulling this if it does >>> overlap. For now, however, until it's clear that something else >>> adequately covers this requirement, I'll keep it in. >> >> Martin? > > The 'wait' tag seems perfectly appropriate to me for monitoring > changes in resource state (aka long -polling). As specified in -04, > it works, though it's not necessarily clear from the text. > > My current thoughts are that you can have a resource with a particular > state, indicated by an ETag. By using If-None-Match (or one of the > other conditional headers) and Prefer: wait=x, then you can request > that the server only provide an update when the resource changes, > within that interval. It's a new use of the conditional headers, as > well as a slightly different spin on the wait header, but I think that > it's workable. > > Of course, you could use the 'wait' tag without anything fancy if the > resource simply had specific logic for long-polling. It seems less > nice that way. > > I might be able to draft a paragraph or two to add if folks are > amenable to this. > > --Martin
Received on Monday, 5 December 2011 23:51:02 UTC