Maybe you could use column_name, dtype, min, max columns and sidestep the nulls, but you’d have to convert “min” and “max” to text so they’re all the same type?

Maybe you could use column_name, dtype, min, max columns and sidestep the nulls, but you’d have to convert “min” and “max” to text so they’re all the same type? Could also split the table by dtype outside of sql to ditch null values without needing to convert non-null ranges to text, but then you split your data into multiple tables, maybe that’s inconvenient to break table schemas into chunks?