Meet DBIish, a Perl 6 Database Interface

In the aftermath of the Oslo Perl 6 hackathon 2012, I have decided to fork and rename MiniDBI. MiniDBI is intended as a compatible port of Perl 5's excellent DBI module to Perl 6. While working on the MiniDBI backends, I noticed that I became more and more unhappy with that. Perl 6 is sufficiently different from Perl 5 to warrant different design decisions in the database interface layer.

Meet DBIish. It started with MiniDBI's code base, but has some substantial deviations from MiniDBI:

• Connection information is passed by named arguments to the driver (instead of a single DSN string)
• Different naming of several methods. There's not much point in having both fetchrow_array and fetchrow_arrayref in Rakudo. fetchrow simply returns an array or a list, and the caller decides what to do with it.
• Backends only need to implement fetchrow and column_names, and get all the other fetching methods (like fetchrow-hash, fetchall-hash) for free.
• Error handling from DB connection and statement handle are unified into a single row

The latter two changes brought quite a reduction in backend code size.

My plans for the future include experimenting with different names and maybe totally different APIs. When a language has lazy lists, one can simply return all rows lazily, instead of encouraging the user to fetch the rows one by one.

Currently the Postgresql and mysql backends support basic CRUD operations, Postgresql with proper prepared statements and placeholders. An SQLite backend is under way, but still needs better support from our native call interface.