The following documentation comment has been logged on the website: Page: https://www.postgresql.org/docs/13/datatype-uuid.html Description:
I'm missing information about what really sets the UUID datatype apart? For example I'm missing information as to how the UUID is sorted, i.e. how two UUIDs are compared to one another. (This is crucial knowledge for a user who tries to understand if the UUID datatype is a good choice, for example as a PK.) The reason why this is not obvious is that there are multiple ways to look at it. Microsoft has written an excellent article on the topic, named "How many ways are there to sort GUIDs? How much time do you have?" [1], which explains it well. So which method does PostgreSQL use? The other thing worth documenting is if PostgreSQL enforces some kind of standard compliance on the UUID. As far as I can tell it doesn't, you can put any 16 bytes into a UUID type and no validation will take place, right?. If that's the case then perhaps worth documenting how the UUID type is really different from bytea(16) - if such a thing existed. Bottom line: from what I can tell, the UUID data type is really just an byte array of length 16 to the backend with no intrinsic knowledge of what those bytes represent. What makes the UUID type special is that - Presentation: When the UUID datatype is presented it is done so in traditional RFC 4122 format, i.e. 'a0eebc99-9c0b-4ef8-bb6d-6bb9bd380a11' - Input: The datatype accepts input in RFC 4122 format ...but apart from this ? Just a byte array, right? /Lars [1] https://devblogs.microsoft.com/oldnewthing/20190426-00/?p=102450
