Re: Push and Caching

On 27 August 2014 05:05, Roy T. Fielding <> wrote:

> > Given the long-lived nature of HTTP/2 connections is this really the
> > right bound on validity?  It's not as though the server is guaranteed
> > an opportunity to replace/invalidate the response.
> I would expect them to push another.  Isn't that sufficient?

I'm with Martin on this one.

Resources are pushed in the context of a GET request, not in the context of
a connection.

Their should be no obligation on the server to monitor the status of the
pushed resource beyond the time it is sent, as that could add up to a
significant active burden on the server to look for file changes in pushed
resources, and other resources may not even be possible to now if they have
changed without attempting to regenerate them.

More importantly, once the initial GET has finished, there is no context in
which to either form the 'fake' request to which the server is pushing a
response, nor a stream on which is can be pushed.

I think the effect of pushing a no-cache resource should be as if the
client had just received a response to a request that it made itself during
the processing of the scoping GET.


Greg Wilkins <> HTTP, SPDY, Websocket server and client that scales  advice and support for jetty and cometd.

Received on Tuesday, 26 August 2014 23:14:32 UTC