Rethinking device renames and name reuse

I'd like to revisit the "rename device" topic. As we could see from numerous requests, people would like to keep a devices name when it has been replaced. I understand that this is not possible in the current scheme of identifying a device solely by its name. So people are always told to add a few changing bits to the name, like a counter, or the current year or whatever.

So why doesn't KB itself add those bits by composing the identifier from two parts, the user visible name and a machine generated (invisible by default) GUID/UUID? Think primary key composed from two columns in an RDBMS.

This way, people could

• reuse the user visible name after device replacement (because the GUID/UUID part has changed)
• and also (pseudo) rename their devices by revoking a device and creating a new one with the user visible part changed and the GUID/UUID kept the same

In both scenarios, the identifier as a whole has changed, which IMHO meets the current requirement of having immutable device identifiers.

Thoughts?

No files, tests, or entry points are identified. Start by reviewing the current device-identifier scheme and the eight-comment discussion; a concrete scope and acceptance criteria are still needed before completion can be assessed.