FeedSync

» see feedsync.org as it lived, on the wayback machine

~ the obituary ~

I have never implemented this and I have never knowingly used software that did, which puts me with almost everybody. FeedSync was a developer specification with no consumer surface at all, and the only way to have met it was to have been building against Microsoft's Live platform between roughly 2007 and 2012.

Ray Ozzie had been at Microsoft about three weeks as chief software architect when he published a post titled "Really Simple Sharing" on 20 November 2005. The observation behind it was narrow and correct. RSS and Atom had become the way items moved around the web, and they moved in one direction only, from a publisher out to subscribers. Add a small block of synchronization metadata to each item and you get a conversation instead of a broadcast: two feeds subscribed to each other, each able to update the other and resolve the conflicts that follow. Ozzie called it a decentralized data bus among the world's websites. The first name was Simple Sharing Extensions, and Dave Winer, who wrote the RSS variants it extended, gave input on the spec.

The rename to FeedSync landed in December 2007 along with a handover: Ozzie's brother Jack and a team at Microsoft took over the maintenance, and feedsync.org went up as a page inside the Windows Live Dev site, sitting in a sidebar between App Storage and Silverlight Streaming. The licensing was deliberately generous for the era, Creative Commons plus the Microsoft Open Specification Promise, because the whole value of a sync format is that other people adopt it.

~ the case file ~

~ where it actually ran ~

Mostly in one place. Windows Live Mesh used FeedSync underneath to keep folders in step across machines, and that was the shipping implementation that mattered. Microsoft announced on 13 December 2012 that Live Mesh would be discontinued, and it went off on 13 February 2013, replaced by SkyDrive, which synchronized files by a completely different route.

From that point the format had no running home. The idea did not die of being wrong, it died of the platform that carried it being replaced by one that did not need it. Sync moved into proprietary services with their own protocols, and the case for a public, feed-shaped way of doing it evaporated along with the assumption that feeds were how data would travel.

~ the spec ~

The last published version is FeedSync for Atom and RSS 1.02, with a specification page dated 3 January 2011. By then feedsync.org had stopped being a site of its own: every capture of it from 2010 onward is a redirect to feedsyncsamples.codeplex.com, a sample project on Microsoft's open source hosting service.

~ the last redirect ~

Microsoft announced on 31 March 2017 that CodePlex was closing. The site went read only in October and went offline on 15 December 2017, taking the sample code and the specification page with it. The redirect at feedsync.org kept pointing at the dead address for a while, which is how a spec can carry on technically existing with nothing at the end of it, and then the domain went too. It does not resolve today.

What is left of FeedSync is the namespace string, written into any feed that ever used it as http://feedsync.org/2007/feedsync, pointing at a host that no longer answers. That is standard for XML namespaces and it is supposed to be fine, since a namespace is only an identifier and was never meant to be fetched. It still reads strangely in a format whose entire purpose was to keep two ends of a connection talking.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »