Validate domain graph before submission

[GillesVanc]

Before sending a domain graph to the user, it would be nice to have some simple validation before actually call the reload endpoint. This because the response of the server does not tell me what is wrong with the domain graph.

Some examples of possible checks:

• no duplicated entities (e.g. detect inconsistencies in the source data)
• no duplicated relations
• only use known entity-types and attribute types (e.g. to detect typos)
• check attribute assignments types
• ...

These are some examples but there are certainly more good checks where the server gives an error but does not specify what is wrong with the reload. Detecting them before reloading would save a lot of time when developing a connector!

[jdeflander]

• no duplicated entities (e.g. detect inconsistencies in the source data)
• no duplicated relations

These two should be easy enough to check, indeed.

• only use known entity-types and attribute types (e.g. to detect typos)
• check attribute assignments types

These two are trickier in the current design, I think. Currently connectors do not have the necessary permissions to retrieve attribute types (nor entity types). Also, these checks provide no real guarantees as the situation may change between checking and actually posting the domain graph. I see two options here:

1. We update our backend so that connectors can retrieve attribute types and entity types. This makes it possible to apply the above checks.

2. We come back to this when the next steps for our connector framework are clear. I actually prefer this option because it seems to me that attribute type management for connectors requires some redesign anyway.

[GillesVanc]

I would indeed choose option 2, for now. That this can then be done in a clean way.

For checking the duplicates in a domain graph: I did write a function to do this my self. Therefore it is also not a high priority for me.