On Wed, Feb 20, 2013 at 11:04 AM, Poluxus <[email protected]>wrote:

> Thanks.
>
> I know the feed i gave you was broken, it was for the example.
>
> Is there a system to tell if a publisher stops using PuSH?
>
How could anyone other than the publisher know that the publisher had
stopped using PubSubHubbub? From a hub's point of view, a feed that is
updated infrequently and one that never updates are largely
indistinguishable. If you were polling the feed, you might notice a removal
of the auto-discovery data in the feed (i.e. <link rel="hub"...),  but that
data isn't required by hubs, it is only there as a hint for potential
subscribers. Thus, removal of the data doesn't really tell you very much;
it could be just that the publisher left it out by accident. The publisher
could continue to notify the hub of new updates even for a feed without
discovery data... A system that tried to deduce when a publisher has
stopped using PubSubHubbub would be trying to "prove a negative." That is
often impossible.


>
> @Jay : I thought that PuSH was created to avoid polling the source

Certainly, a major motivation for PubSubHubbub is the dramatic reduction in
feed polling that results from its use. However, in order to improve the
ease of adoption, it is also intended to a be a fairly simple system and
thus doesn't specifically address all possible user requirements. As
suggested above, being able to notify subscribers when publishers stop
using PubSubHubbub would require additional complexity in the protocol and
would also probably rely on a hard to enforce requirement that publishers
must notify hubs in some manner when they cease to use the hub.

Subscribers who wish to discover when publishers stop using PubSubHubbub
and who are willing to rely on the removal of auto-discovery data in feeds
as an indication of continued hub use, can easily achieve a "compromise"
position by polling feeds to verify the auto-discovery data whenever they
refresh their subscriptions. While such a pattern would still require some
polling, it is likely that it would result in orders of magnitude less
polling than would be done without PubSubHubbub.


>
>  --
>
> ---
> You received this message because you are subscribed to the Google Groups
> "Pubsubhubbub" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"Pubsubhubbub" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.


Reply via email to