News for hackers. Stuff that matters.
Re:New UUID Formats by zbuf on Sunday June 12, 2022 @04:23PM (#31717938)
Thanks for drawing my attention to that, it's the useful answer I was looking for; my use cases haven't been bound by write performance in this manner. However, I'd still be considering carefully before making use of these UUID schemes.
Re:New UUID Formats by jasonwatkinspdx on Sunday June 12, 2022 @06:12PM (#31718760)
If you want something like this, and need a "I just want it to work, require no central coordination, and to have vanishingly small probability of collision" then just use the same concepts but wider than the 128 bit footprint limit of this scheme. This limit makes sense for the RFC in the post, as there backwards compatibility is an explicit goal. But if you used say a 64 bit nanosecond counter (to preserve best case precision on a single machine) along with 128+ bits of random data and you'll need to worry more about gamma ray bursts than collisions.