Struct ColumnarMemtable 

pub struct ColumnarMemtable { /* private fields */ }

In-memory columnar buffer that accumulates INSERTs.

Each column is stored as a typed vector. The memtable flushes to a compressed segment when the row count reaches the threshold.

Implementations

impl ColumnarMemtable

pub fn new(schema: &ColumnarSchema) -> Self

Create a new empty memtable for the given schema.

pub fn with_threshold(schema: &ColumnarSchema, flush_threshold: usize) -> Self

Create with a custom flush threshold.

pub fn append_row(&mut self, values: &[Value]) -> Result<(), ColumnarError>

Append a row of values. Validates types and nullability.

pub fn row_count(&self) -> usize

Number of rows currently buffered.

pub fn should_flush(&self) -> bool

Whether the memtable has reached its flush threshold.

pub fn is_empty(&self) -> bool

Whether the memtable is empty.

pub fn schema(&self) -> &ColumnarSchema

Access the schema.

pub fn columns(&self) -> &[ColumnData]

Access the raw column data (for the segment writer).

pub fn try_dict_encode_columns(&mut self, max_cardinality: u32)

Convert low-cardinality String columns to DictEncoded in-place.

pub fn iter_rows(&self) -> MemtableRowIter<'_>

Iterate rows as Vec<Value>. For scan/read operations.

pub fn get_row(&self, row_idx: usize) -> Option<Vec<Value>>

Get a single row by index as Vec<Value>.

pub fn drain(&mut self) -> (ColumnarSchema, Vec<ColumnData>, usize)

Drain the memtable: return all column data and reset to empty.

pub fn drain_optimized(&mut self) -> (ColumnarSchema, Vec<ColumnData>, usize)

Drain with automatic dictionary encoding for low-cardinality String columns.

pub fn ingest_row_refs( &mut self, values: &[IngestValue<'_>], ) -> Result<(), ColumnarError>

Zero-copy row ingest for timeseries and high-throughput paths.

Accepts borrowed values via IngestValue<'_>, avoiding string cloning for tag columns that are already interned in the DictEncoded dictionary.

pub fn add_column( &mut self, name: String, column_type: ColumnType, nullable: bool, )

Add a new column to the schema, backfilling existing rows with nulls/defaults.