Module writer 

Structs

SeriesIndexRow 🔒

SeriesIndexWriter

Incrementally aggregates sorted flat record batches into series-index Parquet files.

SeriesIndexWriterMetrics

Metrics collected by a SeriesIndexWriter.

SeriesIndexWriterOptions

Options for writing a series index.

Constants

WRITE_BATCH_SIZE 🔒

Functions

decode_primary_key 🔒

Decodes one sparse primary key into a series-index row by extracting only the reserved ids and the tag columns the index needs, instead of decoding every label into owned values.

is_reserved_column 🔒

rows_to_batch 🔒

Builds a batch in the index schema from aggregated rows. The rows carry raw timestamps in unit (guaranteed by timestamp_values), which series_index_schema stamps into the min/max ts columns.

series_index_schema

Returns the Arrow schema of a series index.

tag_columns 🔒

time_index_unit 🔒

Returns the unit of the region’s timestamp time index; the writer stamps it into index files so a searcher interprets each file in the unit it was written with (the region’s time index unit may have been widened since).

timestamp_values 🔒

Extracts the time index column’s raw values in the writer’s unit. A timestamp array carrying a different unit is an invariant violation: the alter path flushes memtables before widening the region’s time index unit, so a writer never receives batches in the region’s previous unit. Reject the mismatch rather than reinterpreting or rescaling the values.

ts_array 🔒

Builds a timestamp array in unit from raw i64 values.

validate_metadata 🔒