On 06/09/2010 01:30 AM, David Recordon wrote:
A few days ago I saw that Feedburner was overwriting TypePad's hub in
my feeds. This occurred even when disabling PingShot.
Conceptually it seems okay and proper that Feedburner, which proxies the
feed content, would have its own hub that proxies the hubbub
notifications rather than linking to TypePad's hub.
(If the FeedBurner feed linked to TypePad's hub directly that would fail
because the TypePad hub allows subscriptions only to feeds recognized as
being on TypePad; the hub will reject subscriptions to a feed that is
served from FeedBurner.)
With everything set up properly, the setup should be:
(End Subscriber)
|
|
(FeedBurner's Hub)
|
|
(FeedBurner's version of the feed)
|
|
(TypePad's Hub)
|
|
(TypePad's version of the feed)
In other words, FeedBurner must act as a PubSubHubbub subscriber in
order to get the real-time updates from TypePad, update its local
representation of the feed and then send out its own ping to its own hub
so that subscribers to FeedBurner's feed will get notified. If
everyone's acting correctly per the above diagram then hubbub will still
work even with the FeedBurner indirection.
I wonder if right now FeedBurner is actually polling the TypePad feed
rather than subscribing to it with hubbub, thus causing polling latency
in the updates to the FeedBurner feed. Can someone at FeedBurner/Google
verify that FeedBurner does actually proxy through PubSubHubbub pings
from its source feeds?
