Not really a JDBC comment, but several of the systems I've worked on actually had a CALENDAR dimension table defined which broke down a given date into various formats, representations and attributes. It was then quite easy to join a fact table on the dimension table via some key field (date_key). In this way we quite commonly (and very easily) would create "faux" records via SQL if there was no data for a given date in the fact table.
I should note, these were ETL and reporting data warehouse systems so whether or not you do something like that is really more requirement/system specific. For these, more often than not the final output of most processes and external entities accessing the data needed to know whether or not there was data for a given day and what those values were, even if zero. This approach centralized the issue so that simliar logic was not needed in various external, unrelated systems.