HTTP rules? (was: [hybi] [Uri-review] ws: and wss: schemes)

David Booth wrote:
> If the Tachyon Protocol does become dominant in 2272 it will be a simple
> matter to define a new URI scheme, so that Tachyon URIs can be shortened
> henceforth.  (That's the *easy* part.  The hard part was convincing the
> 186,000 makers of nearly 4.2 trillion devices to *implement* the Tachyon
> Protocol.)  But until that point, there is greater value in layering the
> Tachyon Protocol on top of the dominant protocol, whether that happens
> to be HTTP or the InterPlanetaryProtocol.

I was going to let this debate run its course, as these things often do, but 
your comment has sparked an interesting possibility...

The argument to date that I've heard for using HTTP for identifiers is that the 
URI can point to a description of the thing identified.  But your comment hints 
at the possibility of extending this to machine processing.

The notion is that an HTTP URI can also retrieve a protocol hander 
implementation of (say) the tachyon protocol which can be loaded and activated 
to handle the remaining part of the URI.  Allowing browsers to seamlessly handle 
future protocol designs seems like a fitting use for a tachyon protocol :)

I wonder if this might not be done today for HTTP-layered protocols like AtomPub?


Received on Friday, 11 September 2009 05:44:38 UTC