About UUID v7 implementation

[Ai Hoshino]

This function is used in Uuid::new_v7(), it will place the content of rand_bytes[1](12 bits) before rand_bytes[0].

uuid/src/timestamp.rs

Line 235 in 6188ecf

 (random_bytes[0] as u16 | ((random_bytes[1] as u16) << 8) & 0x0FFF) | (0x7 << 12); 

Currently, it has no impact on the generated result because they are both random bytes. Maybe this line should be changed to

diff --git a/src/timestamp.rs b/src/timestamp.rs
index e49771d..7ec88e6 100644
--- a/src/timestamp.rs
+++ b/src/timestamp.rs
@@ -232,7 +232,7 @@ pub(crate) const fn encode_unix_timestamp_millis(millis: u64, random_bytes: &[u8
     let millis_low = (millis & 0xFFFF) as u16;
 
     let random_and_version =
-        (random_bytes[0] as u16 | ((random_bytes[1] as u16) << 8) & 0x0FFF) | (0x7 << 12);
+        (random_bytes[1] as u16 | ((random_bytes[0] as u16) << 8) & 0x0FFF) | (0x7 << 12);
 
     let mut d4 = [0; 8];

Ref: https://datatracker.ietf.org/doc/html/draft-peabody-dispatch-new-uuid-format-04#section-5.2

[Ashley Mannix]

Well spotted! I don’t think there’s any reason why it should be one way or the other, but the fact that the indexes are reversed might suggest there’s more going on than there actually is.

I’d happily accept a PR that reversed them 🙂