Since fields are scanned for types on a row-by-row basis, it can happen that a string value is mistaken for a numeric (e.g. with values such as '1234', '123x', ...). This creates a problem as inserting values of different types is not supported. This option would simply insert all fields as strings, with the exception of timestamps.
A better solution might have been to have the possibility to specify the column types, or to infer them based on a subset of the CSV, rather than independently for each row. But implementing these would involve more effort.
Since fields are scanned for types on a row-by-row basis, it can happen that a string value is mistaken for a numeric (e.g. with values such as '1234', '123x', ...). This creates a problem as inserting values of different types is not supported. This option would simply insert all fields as strings, with the exception of timestamps.
A better solution might have been to have the possibility to specify the column types, or to infer them based on a subset of the CSV, rather than independently for each row. But implementing these would involve more effort.