Information Objects UUID definition good?

[danielmarschall]

Previously, OIDplus has used UUIDv3 and UUIDv5 with custom namespace as OIDplus Information Objects.

Recently, this was changed to custom UUIDv8.

I begin to feel uncomfortable with this solution, though, since these UUIDs can collide easily. On the other hand, it is nice that you have so many data in the UUID that can be extracted, like SystemID and Timestamp...

Should we rather do UUIDv3, UUIDv5 or a Name-Based-UUIDv8 instead? (With a new namespaces to define the different types to be identified)

[wehowski]

• Version 3 and 5 may be in use yet somewhere? So it would be not good to remove it?
• As OID**plus** OIDplus should add Version 8 additionally?
• [s]Maybe we could tag/title/namespace the UUIDs/altIDs by usecase[/s] - We can extract the version from the UUID

[wehowski]

Hello @danielmarschall ,
one other idea could be to (optionally) track the Information Objects UUIDs/AltIds statically in a database table, so that later changes to the system would not affect existing relations?
EDIT: the special object type "ns" was a try to an (advanced custom) object type something like "relation".