WIP: Mappings

[mferrera]

From discussion 11 Feb 2025

[perolavsvendsen]

Summary of the attempted point on changing the current workflow around data.name and SMDA-linking.

Concepts:

• "FMU name": The name used inside FMU (also de-facto an identifier)
• "SMDA name": The name used in SMDA (also de-facto an identifier in combination with the strat column)

Today:

• General rule: data.name is the "FMU name" when exported
• Exception: When data.name is referring to an instance which is also defined in the master data:
  • data.name is changed to the SMDA name.
  • data.is_stratigraphic is set to True.
  • The FMU name is appended to the list of alias.

The end result of this is that, for data objects that are instances of an entity defined in master data, the data.name equals the master data name - which is the identifier.

This is stuff we have inherited from the definitions of metadata made before fmu-dataio became a thing.

Confusing much?

Possible better solution to this:

• All data objects are exported with their "FMU name". No exceptions.
• For data objects which are instances of some entity defined in master data (stratigraphy, well, etc), we add reference in the metadata, something like:

data:
  name: <FMU name, unchanged>
masterdata:
  smda:
    field: ...
    country: ...
    something: # <--
      name: ...
      uuid: ...
      whatever: ...

• The presence of this is the equivalent of the current data.is_stratigraphic

(This may be a derailing of the mapping discussion, but feels like fixing this would make it easier to do mapping back and forth.)