Data Virtualization - Import Endpoint Schema from Existing Dataset

Hi ​@chriswilson ! To make sure that I’m understanding, are you imagining something akin to the “Import > From Dataset...” feature that you can leverage to populate the values for a Choice User Parameter in FME Workbench?

We have considered this idea more in the context of importing schema from an OpenAPI specification. For example, the ability to select a specific schema from an OpenAPI JSON file to add to your existing API—instead of the current behaviour where you can import “all or nothing”.

I’ll speak to a few upcoming features in the upcoming Data Virtualization in Action webinar on Wednesday (August 27th 2025), but we have not yet implemented this ability or an equivalent.

Thanks ​@rylanatsafe,

That’s just about it, though I use that feature in Form more for setting up writers/FeatureWriters e.g. ArcGIS Online feature layers when I want to update an existing dataset.  An auto-import of schema that saves time and error in creating schema manually.

It doesn’t have to be all or nothing, you could choose attributes to import via a wizard, but if we want to present back the data we are going to be reading/processing through the workspace - e.g. a straight-up data request - then importing and having the schema preserved/set correctly would be a good option.  For a large schema if everything was needed this would be borderline if not essential, and yet you could still select certain attributes to simplify your output schema.

Of course the OpenAPI spec is good, but it doesn’t cover the specifics of a given dataset’s schema, and so I think this functionality would compliment that while still staying within the stucture/framework of an OpenAPI-compliant spec.

Will check out the webinar when it comes out.

I started adding this as an idea myself when I got this post as a suggestion.

I miss something a bit like ‘Import Parameters from JSON’ in the Output attributes of automation triggers, or the ‘import from dataset’ option in a FeatureWriter.

Something to aid getting most of the structure right in more complex datamodels.

This idea is much needed! I have a database tables with 50+ attributes and going through adding a schema one attribute at a time is mind numbing

As a workaround for this omission of being able to import schemas, I have been using json variables and doing what I need to do with columns in the workspace. For example, in a POST endpoint the Request.Body Structure > Create Properties > Name: new_data, Type: Array<Object>. When new data is received by the workspace I JSONFlatten and JSONFragment the json data and expose the attributes I need.

For a GET endpoint I set an HTTP Status Code: 200 with Content Type: application/json and Reponse Structure > Create Properties > Name: return_data, Type: Array<Object>. In the workspace for this endpoint, SQLExecutor gets the data and JSONTemplater formats the return data as a json type to the client.

For endpoints which deliver fewer than 5 columns/attributes, I will define a schema still but otherwise I send/receive json data.