Just occurred to me reading the NAIAD stuff

[Craig Latta]

> How do you do your UUIDs? ...Are your UUIDs deterministic?

     RFC 1422 (http://www.ietf.org/rfc/rfc4122.txt), currently type 4
(pseduo-random) while I work on more important things. There's nothing
preventing using any of the other four types (time-based, POSIX, MD5,
SHA-1). I also assert the probability of collisions with any of the five
types is sufficiently small, and overwhelmingly so, given the additional
information combined to create the actual IDs used in the system. My
only requirement so far is uniqueness (nothing with regard to sorting,
etc.).

> If you are using SHA UUIDs it may be easy to implement a versioning
> object store similar to Git or Venti.

     Sure. Naiad is already a versioning object store, although geared
toward a specific application (recording development history). One could
certainly generalize it.


-C

--
Craig Latta
www.netjam.org/resume
+31   6 2757 7177 (SMS ok)
+ 1 415	 287 3547 (no SMS)