* Use r_mem_pool
* Add support for one level RDB.. to 8x speed up
* Serialize/Deserialize from disk
* Return FALSE when adding two elements with the same
  attributes (get before add) if the db is unique
  // not necessary, we can just do this by manually checkingc
* Benchmarking
  - Compilation support for profiling memory usage
  - Compare performance against sqlite
* Import/export from/to SQL
* SQL support

CREATE TABLE flags (
  name char(25) PRIMARY KEY,
  offset ut64,
);

SELECT * FROM flags WHERE offset == 0x8048000;

CREATE VIEW flagwindow AS SELECT * FROM flags WHERE offset>0 and offset<100

DROP TABLE flags
INSERT INTO flags (name, offset) VALUES ("patata", 33)

UPDATE flags SET offset=33 WHERE offset==20

STORE
  - ut32
  - ut64
  - Endianness matterz
  - RList *get_range (0, 4096)
