Google I/O, Salmon, and the Open Web

Last week was intense, and I'm just now coming up from air from the backlog.  We launched a huge expansion to the Buzz API at Google I/O.  My contribution was to ensure that PubSubHubbub real time updates flow for the new feed URLs as well as the older ones; as part of this, we also enabled "fat pings" to the hub from Buzz.  "Fat pings" in this case means that we're doing an active push of each full update from our back end systems through to the hub, so the hub never needs to call back to the feeds to retrieve updates.  This is a more complicated approach, but reduces the overall server load and makes consistency guarantees easier for globally distributed systems.

Just before Google I/O, I went to IIW10 and talked about Salmon, LRDD, Webfinger, and OpenID Connect.  The air at IIW was thick with new specifications and updates.  I would've liked to have participated more, but I was helping to roll out PubSubHubbub for Buzz and getting ready for the "Bridging the Islands" Google I/O talk I was to give with Joseph Smarr the next day.

And then, my laptop crashed.  I blame Chris Messina -- Joseph and I were stealing some of his slides, and I was editing his slide deck when the system froze up and refused to boot.  Turned out later that accessing that file somehow corrupted the disk encryption driver.  That's right, Chris Messina's slide decks are powerful enough to destroy laptops.

Fortunately, we recovered and had a great session (some audience notes here).  Here's a screencast of the cross site Salmon mention demonstrated in the session.  This shows a user on Cliqset mentioning a couple of buddies using other services (StatusNet and a scrappy startup called Atollia), having information flow between the sites, and having mentions be pushed the other services.  The big take-away here is that these services don't need to pre-register or federate with each other any more than they need to federate to send email back and forth; it all Just Works.