Degree of time-sort sensitivity

[Ceyword]

Hello,
1- Can I rely on this implementation of ULID to always accurately generate ULIDs according to time i.e. if I generate many within the same millisecond, from different machines (in a distributed fashion) will they all be accurately time-sorted?
2- If the answer is yes, will the string format achieve the same time-sort sensitivity? The plan is to generate them, then store in a database as strings, to be time-sorted.
Thank you in advance.

[neuecc]

There is no guarantee when crossing between machines, or even on the same machine when IDs are generated at truly identical timestamps.
Handling these cases correctly is a pipe dream within Ulid's specifications.
For proper timestamp handling, you should read the papers on Spanner and TiDB.
In any case, this is beyond the scope of this library.