Provide a parameter for Date/Time format

[csv2db]

Provide a parameter for Date/Time format

[gvenzl]

It is unclear whether that feature provides significant value to be implemented. Users today can load data into a staging table with all columns being of a string data type and then perform the data type conversion inside the database.

There is a valid argument that dates should be converted correctly when loaded into a table with the correct data types, hence requiring a way of passing on the date format (as NLS value for example with Oracle). However, the question is where to start and where to stop. For example, what to do with TIMESTAMP columns, and what to do with NUMBER columns in different regions in the world where the fraction symbol is , and not .. What about character set conversion, etc.

By having date format available to be passed on as command line parameter, users could be lead to expect all other conversions to be provided as well, quickly moving the complexity of data type conversion out from the database and into csv2db and further increase the already long list of parameters. Since the very inception of csv2db it has been a core principle for it not to become an ETL tools with all the bells and whistles. There are already many, very good ETL tools out there and the world doesn't need yet another. Instead, csv2db should aid users as a simple command line tool to get rows from a delimited file into a database table, and not more! According to these principles, the question of data type conversion can be answered by the user in three ways:

1. Load the data into a staging table with all string columns, then perform the data type conversion with database provided functionality such as INSERT INTO SELECT and/or stored procedures. However, this does not solve the question of character set conversion.
2. Have the database configured correctly to match the territory and incoming data of the files, hence allowing for implicit data type and avoiding unnecessary character set conversion
3. Do not use csv2db but one of the many, good ETL tools instead.

[gvenzl]

This ER will not be implemented based on the feedback above.