Details
-
Task
-
Resolution: Won't Do
-
Minor
-
Cheshire-Cat
-
None
-
1
Description
The datastore interfaces for namespaces, buckets, scopes and keyspaces have the ability to access the underlying objects by Id and by name (eg KeyspaceNames() and KeyspaceIds() or KeyspaceByName() and KeyspacebyId()), and have corresponding Name() and Id() identifiers.
While for indexes the IndexId is a proper UUID, for datastore the Id and the name are the same, and the n1ql code is quite brazen as to whether it uses KeyspaceByName() or KeyspaceById() etc.
While this is harmless, it's messy, and it requires maintaining two ids for no reason whatsoever.
I propose that we get rid of all the datastore objects Id(), ...Ids() and ...ById() variants.