When the format is set to JSON, and Content Type is set to application/json, the individual records are saved as JSON objects, but are not comma-separated, and aren't in a JSON array, so the file itself is not a valid JSON file.
Actual file output:
Expected File Output:
This is the intended behavior, it’s not technically a valid json element, but this is the format that the source reads and that many other distributed systems work with (for example, you can load this file directly to BigQuery). The reason has to do with being able to split the file in order to read it efficiently. If it is a proper json array, there isn’t a way to split it.
That being said, it would be worthwhile to add an option to write it out as proper json, assuming that is what is wanted by the user.