+1

It's no longer a necessity. When we cooked up the Zend library, we realised 
that just by adding a subscription store we could handle both cases very 
easily. The introduction of automatic refreshes just confirmed that async would 
be the subscriber standard.

Paddy

 Pádraic Brady

http://blog.astrumfutura.com
http://www.survivethedeepend.com
OpenID Europe Foundation Irish Representative





________________________________
From: Blaine Cook <[email protected]>
To: pubsubhubbub <[email protected]>
Sent: Thu, March 4, 2010 12:10:03 AM
Subject: [pubsubhubbub] Removing Async / Sync

Hi all,

Mike, Alexis, and I had a very productive Thursday in London
discussing the finer points of the PSHB spec; herewith a few proposals
to improve the specification and hopefully enable some functionality
that many of us are interested in.

The first proposal is to remove the sync/async hub.verify parameter;
chatting with Brett, it seems that this parameter used to be
meaningful, but with the addition of automatic subscription
refreshing, subscribers must now effectively support the async mode,
and all subscribers must effectively support the sync mode in order to
avoid timing issues.

The consensus in London was that removing the distinction simplifies
the spec, simplifies the implementation of hubs, and has virtually no
impact on subscribers. In addition, it has the positive side-effect of
ensuring that subscribers must understand a common and consistent set
of subscription response codes; more on that to follow.

It's worth noting that this change is backwards-compatible, in that
hubs can safely ignore the hub.verify parameter so long as they employ
synchronous verification in all cases (which is what the reference hub
does already), with the slight modification that subscribers may
receive 202 Accepted responses in reply to verification requests at
any time.

A diff against Version 0.3 of the PSHB spec, and the complete modified
spec is included.

b.

Reply via email to