Struct Database 

pub struct Database { /* private fields */ }

Implementations

impl Database

pub fn set_spill_threshold(&self, bytes: u64)

Override the spill threshold at runtime (via SET spill_threshold). A value of 0 explicitly disables spilling.

pub fn effective_spill_threshold(&self) -> u64

Get the effective spill threshold in bytes.

Priority:

1. Runtime override (via SET spill_threshold; 0 = explicitly disabled)
2. config.spill_threshold
3. 80% of buffer_pool_size
4. 0 (disabled)

pub fn spiller(&self) -> Option<Arc<Spiller>>

Create a Spiller instance using the current database configuration.

Returns None if spilling is disabled (threshold is 0).

pub fn storage_driver(&self) -> StorageDriver

Create a [StorageDriver] for programmatic storage-level access (page counts, buffer stats, file sizes, table counts) without Cypher.

pub fn catalog(&self) -> Arc<Mutex<Catalog>>

Get a reference to the schema catalog for programmatic metadata access.

pub fn table_catalog(&self) -> Arc<TableCatalog>

Get the data table catalog for programmatic data access.

Prefer using the unified DDL methods on Database instead of accessing the table catalog directly.

pub fn create_node_table( &self, name: String, columns: Vec<CatalogColumn>, ) -> Result<u64, String>

Create a node table: data table + serial sequences + ART index.

The schema entry is created by the binder during bind(). This method creates the storage-level table and associated resources.

pub fn create_rel_table( &self, name: String, src_table_id: u64, dst_table_id: u64, columns: Vec<CatalogColumn>, ) -> Result<u64, String>

Create a rel table: data table.

The schema entry is created by the binder during bind(). This method creates the storage-level table.

pub fn drop_table(&self, name: &str) -> Result<(), String>

Drop a table: serial sequences + data table + schema entry.

pub fn refresh_vector_indexes(&self, _table_ids: &[u64])

No-op when the vector extension is not compiled in.

pub fn create_art_index( &self, table_name: &str, index_name: &str, ) -> Result<(), String>

Create an ART index on a node table: data index.

The schema entry is created by the binder during bind().

pub fn drop_art_index( &self, table_name: &str, _index_name: &str, ) -> Result<(), String>

Drop an ART index from a node table: data index + schema entry.

pub fn table_num_rows(&self, name: &str) -> u64

Get the number of rows in a table by name.

pub fn get_table_id(&self, name: &str) -> Result<Option<u64>, String>

Get table IDs for write operations (used by transaction locking).

Propagates a poisoned catalog lock instead of silently treating it as a missing table (Audit 2 NIT).

pub fn catalog_file_path(&self) -> PathBuf

Path of the persisted catalog file for this database.

pub fn connect_tcp(addr: impl Into<String>) -> Result<RemoteDatabase, String>

Connect to a remote Akar server (embedded server mode, P47).

The server process owns the Database instance and its exclusive file lock; remote clients only talk to the server over TCP and never open the database directory themselves. Multiple processes can therefore access the same database concurrently: one writer via the server plus any number of read-only clients (and additional writers when concurrent_writes is enabled).

See crate::remote::RemoteDatabase for the returned handle’s API.

pub fn is_in_memory(&self) -> bool

Returns true when the database runs fully in-memory (:memory:), in which case catalog persistence is skipped.

pub fn persist_catalog(&self) -> Result<(), String>

Persist the system catalog to disk.

Called after every DDL statement so schema changes survive restarts. The write is atomic (temp file + rename) and no-ops for :memory: databases.

pub fn new( db_path: impl Into<PathBuf>, config: SystemConfig, ) -> Result<Self, String>

Open or create a database at the given path.

Arguments

• db_path — filesystem path where database files are stored.
• config — buffer pool size, thread count, etc.

Errors

Returns Err if the path is not writable or if existing data is corrupt.