Ok, so, I have taken everyone's concern into account and I have forked the spec so that it looks like that now: https://github.com/pubsubhubbub/PubSubHubbub/commit/9a2455c68aa495691179295353dfe00bef287019
Would this make everyone happy? Generally, I tried to "ease" the relationship between hubs and publisher and leave room for other custom mechanism (or no mechanism at all), while keeping the current mechanism (POST with hub.mode=publish, and fetch) for the default use case. Julien On Mon, Nov 28, 2011 at 3:45 AM, Monica Keller <[email protected]>wrote: > >>I think that the specification should assume the most decentralized > case, as this is the broadest and the most beneficial case on the open Web. > > The problem is that this is not the most common case. It was not the case > for MySpace or for Facebook or for any API provider whose apis require > authentication. They cannot fanout and syndicate the same response to more > than one consumer. > > IMO the most important thing is to promote event driven PuSH architecture > as opposed to continuous polling. >
