We've just released a new db-dtypes package and would like to hear your thoughts.
This package introduces two new extension dtypes to pandas: "time" and "date"
Goals of the package include:
- Provide a natural mapping to/from database-like systems such as BigQuery, which provide a TIME and DATE data type.
- Natural mapping to/from existing conventions for date and time data. For example, a common convention for date columns is to represent them as datetime64 values @ midnight. Also, it's common to represent time columns as datetime.time objects with object dtype (such as is done in read_sql).
- It should also be possible to combine data and time values to create datetime values.
Please take these for a spin and let us know if they don't meet your expectations by filing an issue on GitHub
Thanks for you help!