6638 added support to scan multiple JSONL files, but not for multiple JSON files AFAICT. This would also be super useful.
While reading multiple files, it would be helpful in general to (optionally) have an automatic/virtual column of file metadata associated with each row, so we can tell which row came from where. The most useful metadata would be path & modified time IMO. (Duckdb, clickhouse can both add the filename, neither can do modified time).
wasimsandhu
commented
3 weeks ago
Description
6638 added support to scan multiple JSONL files, but not for multiple JSON files AFAICT. This would also be super useful.
While reading multiple files, it would be helpful in general to (optionally) have an automatic/virtual column of file metadata associated with each row, so we can tell which row came from where. The most useful metadata would be path & modified time IMO. (Duckdb, clickhouse can both add the filename, neither can do modified time).