Struct ColumnMeta 

pub struct ColumnMeta {
    pub name: String,
    pub data_type: u8,
    pub role: u8,
    pub default_encoding: u8,
    pub stats: ColumnStats,
    pub bloom_filter: Option<Vec<u8>>,
    pub encrypted: bool,
    pub key_id: Option<String>,
    pub decimal_scale: Option<u8>,
    pub row_group_blooms: Option<Vec<Vec<u8>>>,
}

Metadata for a single column within a segment.

Segment-level statistics

The stats field stores aggregate statistics across all row groups: min_value, max_value, null_count, value_count, sum, and distinct_count. These are computed by the segment writer during encoding and merged via merge_stats(). Query-time code can use these to prune entire segments without reading row-group data.

Metadata is kept proportional to the data: no fixed-size sketches or histograms, which cost ~49 KiB per segment however little it holds, and which the planner does not need — distinct_count above is exact.

Fields

name: String

Column name.

data_type: u8

Column data type tag (maps to ColumnType).

role: u8

Column role tag: 0 = timestamp, 1 = tag, 2 = field.

default_encoding: u8

The encoding the writer intended for this column, chosen from the column’s type before any data was seen.

This is a hint only. The adaptive selector may fall back to a different codec per block (e.g. Chimp → Gorilla → plain), so the authoritative encoding for reading is always ColumnBlockMeta::encoding. The field was previously named encoding, which read as a statement of fact; nothing decodes from it, but the name invited that mistake.

stats: ColumnStats

Global statistics across all row groups (min, max, null_count, value_count, sum, distinct_count). Enables segment-level pruning at query time.

bloom_filter: Option<Vec<u8>>

Bloom filter for tag columns.

Contains the serialized bloom filter bit array for tag columns, enabling segment-level predicate pushdown: if a queried tag value is definitely not in the bloom filter, the segment can be skipped. None for non-tag columns, and for segments written before per-tag row-group blooms existed.

encrypted: bool

Whether this column is encrypted at rest (field-level encryption).

When true, all column blocks for this column are AES-256-GCM encrypted and key_id identifies the encryption key. Statistics (min/max/sum/distinct) are zeroed to prevent leakage.

key_id: Option<String>

Encryption key ID for this column. Used by the reader to look up the decryption key from a FieldKeyProvider.

decimal_scale: Option<u8>

Digits after the decimal point, for a decimal column.

None for every other type. The scale lives here, once per column, rather than beside each value: it is a property of the column, fixed when the column was created, and the write path rescales every value to it before the memtable ever sees it. Storing it per value would cost a byte a row to say the same thing 65 536 times, and would make a segment whose rows disagreed representable.

row_group_blooms: Option<Vec<Vec<u8>>>

Per-row-group bloom filters for string columns .

row_group_blooms[rg_idx] is a serialized bloom filter containing all distinct string values for this column in that row group. Enables row-group-level predicate pushdown: if a queried value is definitely not in the bloom filter, the row group can be skipped. None for non-string columns, encrypted columns, or segments written before .

Implementations

impl ColumnMeta

pub fn arrow_data_type(&self) -> DataType

The Arrow type this column decodes to.

The single answer to a question four places in the reader used to answer separately, each ending in _ => DataType::Utf8. That fallback was fine while every non-numeric column was a string; it silently turned a decimal column into Utf8 the moment one existed.

pub fn empty_array(&self) -> ArrayRef

An empty array of this column’s Arrow type, for a pruned row group.

pub fn arrow_field(&self) -> Field

The Arrow field for this column, carrying its role as metadata.

Trait Implementations

impl Clone for ColumnMeta

fn clone(&self) -> ColumnMeta

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ColumnMeta

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ColumnMeta

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for ColumnMeta

fn eq(&self, other: &ColumnMeta) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for ColumnMeta

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for ColumnMeta