On 1/24/11 8:36 AM, "Simon Slavin" <slav...@bigfraud.org> wrote:
> > On 24 Jan 2011, at 4:21pm, Duquette, William H (318K) wrote: > >> A question on using randomblob(16) to generate UUIDs, as the SQLite docs >> suggest: what assurance do you have that the UUID really is universally >> unique? It's a pseudo-random number, and you can replicate a stream of >> pseudo-random numbers by setting the seed appropriately. Is randomblob() >> doing some kind of magic in its seeding of the random number stream? > > Your assurance is only statistical. Version 4 UUIDs have 30 4-bit higits and > one 2-bit higit. That gives you > > (30 * 4) + 2 == 122 > > bits of randomness, which is about 5e36 different numbers. You can work out > yourself how fast people would have to choose random numbers to stand a chance > of one duplication in ten years. Provided that your starting seed is chosen in a sufficiently random way, which evidently it is. A bad choice of starting seed could bring the whole thing crashing to the ground. > > > Simon. > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users -- Will Duquette -- william.h.duque...@jpl.nasa.gov Athena Development Lead -- Jet Propulsion Laboratory "It's amazing what you can do with the right tools." _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
