(In this message, where ever it says "google reader" it can be any
subscriber)

Here's a scenario:
1. I know Google Reader (for instance) has the capabilities to be a
subscriber.
2. I use Google Reader to subscribe to a feed hosted at my server,
this feed states the hub to be my server.
3. I get a subscription request from Google Reader, this enabling me
to discover their callback URL.
4. I figure out the callback URL of a different atom feed people are
subscribed to that has a different hub, Then I send a POST request to
that callback with new "fake" content causing Google Reader's users to
see a fake post..

So yeah, If google use a custom callback url they can easily use
something like this for example:
http://reader.google.com/pushCallBack/feedid/[feed id here]/hash/[hash
key for feed here]
Then when getting a POST request they can check the hash actually
matches what they expect, and hashing can be done with a secret key or
whatever.

Simple, Right?
However, that would totally scratch the possibility of using the
"Aggregated Content Distribution" stated in the protocol:
http://pubsubhubbub.googlecode.com/svn/trunk/pubsubhubbub-core-0.2.html#aggregatedistribution
Which basically says "different callbacks generate different POST
requests and no aggregation"

I'm surprised a way for a hub to identify itself to a subscriber when
POSTing new content isn't in the protocol.
Or am I missing something?

(If there's another post dealing with this issue, I'd love to be
pointed to it, a quick search didn't help..)

Thanks for the help :)

Reply via email to