Does the system clock guarantee in-order responses?

[danlentz]

Just asking out of curiosity, as when working updating clj-uuid for rfc-9562 I noticed that, even in a single thread, I could generate enough calls that I would eventually receive out-of-order clock times.

uuid/uuid.lisp

Line 143 in f0052f3

(cond ((not (= last-time time-now))

if so, mitigating this would require just a small additional clause:

((> last-time time-now)
  (sleep 0.0001)
  (go restart))

Similarly to here https://github.com/danlentz/clj-uuid/blob/master/src/clj_uuid/clock.clj#L68

Cheers!

[danlentz]

well, ok, the sleep 0.0001 in your loop probably helps avoid this happening in practice for single threaded. (and we're not worrying about multiple threads here). But, anyways, just passing along the experience, as it was a tricky one to figure out.