Trait kvdb::KeyValueDB

source · [−]

pub trait KeyValueDB: Sync + Send {
    fn get(&self, col: Option<u32>, key: &[u8]) -> Result<Option<DBValue>>;
    fn get_by_prefix(
        &self, 
        col: Option<u32>, 
        prefix: &[u8]
    ) -> Option<Box<[u8]>>;
    fn write_buffered(&self, transaction: DBTransaction);
    fn flush(&self) -> Result<()>;
    fn iter<'a>(
        &'a self, 
        col: Option<u32>
    ) -> Box<dyn Iterator<Item = (Box<[u8]>, Box<[u8]>)> + 'a>;
    fn iter_from_prefix<'a>(
        &'a self, 
        col: Option<u32>, 
        prefix: &'a [u8]
    ) -> Box<dyn Iterator<Item = (Box<[u8]>, Box<[u8]>)> + 'a>;
    fn restore(&self, new_db: &str) -> Result<()>;

    fn transaction(&self) -> DBTransaction { ... }
    fn write(&self, transaction: DBTransaction) -> Result<()> { ... }
}

Expand description

Generic key-value database.

This makes a distinction between “buffered” and “flushed” values. Values which have been written can always be read, but may be present in an in-memory buffer. Values which have been flushed have been moved to backing storage, like a RocksDB instance. There are certain operations which are only guaranteed to operate on flushed data and not buffered, although implementations may differ in this regard.

The contents of an interior buffer may be explicitly flushed using the flush method.

The KeyValueDB also deals in “column families”, which can be thought of as distinct stores within a database. Keys written in one column family will not be accessible from any other. The number of column families must be specified at initialization, with a differing interface for each database. The None argument in place of a column index is always supported.

The API laid out here, along with the Sync bound implies interior synchronization for implementation.