CLI: Add support for metadata import

[ptitzler]

Is your feature request related to a problem? Please describe.
With the [planned] delivery of #2448 in Elyra 3.7, users can export metadata for individual schemaspaces using the CLI:

elyra-metadata export <schemaspace>  ...

Elyra should also support import of exported metadata

elyra-metadata import <schemaspace> ...

Describe the solution you'd like

Constraints (based on initial discussion in #2448):

• a user should be required to provide only the minimum number of input parameters, e.g. there should be no need to specify a schema_name
• only valid schemaspace instances can be imported (export supports export of invalid instances)
• only non-deprecated schemaspace instances can be imported (export supports export of deprecated instances)
• a user should be able to specify what should happen if a conflict is detected, e.g. if an instance with the same name is already present (default behavior: tbd)

[kevin-bates]

Thanks for opening this issue @ptitzler. I have a couple of comments and questions regarding the constraints.

only valid schemaspace instances can be imported (export supports export of invalid instances)

Should the behavior be that attempts to import invalid instances result in a message, that instance is skipped, and processing continues? I would prefer that the invalid instance "creation" be attempted, otherwise, it implies that validation of the invalid instance would need to occur before the import attempt.

a user should be able to specify what should happen if a conflict is detected, e.g. if an instance with the same name is already present (default behavior: tbd)

I think the default behavior must be that the importing-instance be skipped if an instance of that name exists. Seems like --overwrite might be a reasonable flag to force an update (actually, perhaps --force is better).

[ptitzler]

Thanks for opening this issue @ptitzler. I have a couple of comments and questions regarding the constraints.

only valid schemaspace instances can be imported (export supports export of invalid instances)

Should the behavior be that attempts to import invalid instances result in a message, that instance is skipped, and processing continues? I would prefer that the invalid instance "creation" be attempted, otherwise, it implies that validation of the invalid instance would need to occur before the import attempt.

+1

• if possible, validation should be performed by the server (the CLI only submits the requests to add the instances)
• processing should continue after an instance was rejected (otherwise a user would have to remove the invalid instance from the local file system to make progress)

a user should be able to specify what should happen if a conflict is detected, e.g. if an instance with the same name is already present (default behavior: tbd)

I think the default behavior must be that the importing-instance be skipped if an instance of that name exists. Seems like --overwrite might be a reasonable flag to force an update (actually, perhaps --force is better).

I agree about the default behavior. --force appears to be most commonly used flag for this kind of scenario, thus +1 from me.

[ptitzler]

I think the default behavior must be that the importing-instance be skipped if an instance of that name exists. Seems like --overwrite might be a reasonable flag to force an update (actually, perhaps --force is better).

Sorry about flip-flopping here, but --replace would perhaps be a better option name:

• it'd consistent with the existing flag for updating existing instances (elyra-metadata install <namespace> ... --replace)
• is less ambiguous, because force can also be interpreted as 'import invalid metadata instances'

[kevin-bates]

Those are great points - I agree on both counts - thanks!