[T-hawk]

If you're going to use a key space in the trillions, and partitioned and sparsely populated rather than sequential... isn't that just reinventing what the UUID already is?

A UUID is just a 128-bit integer, with creation algorithms designed to partition that space by things that already have enough entropy to need no further synchronization.

What you're proposing sounds like roll-your-own-UUID, which might be similarly inadvisable as roll-your-own-crypto.

[andrewflnr]

Don't try to compare this to rolling your own crypto. The stakes are nowhere near the same for IDs as for crypto, and the stakes are the defining feature of the "don't roll your own crypto" meme.

[T-hawk]

Yes I'm comparing it and you don't get to tell me I can't.

It's not about the stakes. It's the idea that in rolling your own, you're going to get it wrong, or otherwise do worse than existing ways that have already solved the same problem.

[andrewflnr]

That applies to literally all code. So if it's not about the stakes, only experts are ever allowed to deploy code of any kind that they've written themselves.