Struct MetadataPartitionSpec 

pub struct MetadataPartitionSpec {
    pub view_metadata: ViewMetadata,
    pub schema: Arc<Schema>,
    pub insert_range: TimeRange,
    pub record_count: i64,
    pub data_sql: Arc<String>,
    pub default_audience: Option<Arc<str>>,
    pub compute_time_bounds: Arc<dyn DataFrameTimeBounds>,
    pub sort_order: Option<Vec<String>>,
    pub unfiltered_count: Option<i64>,
}

Fields

view_metadata: ViewMetadata

schema: Arc<Schema>

insert_range: TimeRange

record_count: i64

data_sql: Arc<String>

default_audience: Option<Arc<str>>

Bound as $3 when data_sql runs – the deployment’s MICROMEGAS_DEFAULT_AUDIENCE, which BlocksView’s COALESCE resolves a never-stamped process’s missing audience to. None for a data_sql that references only $1/$2; BlocksView is the only view that uses this module, so in practice it is always Some. Note the separate source_count_query deliberately does not get this bind – it has no $3.

compute_time_bounds: Arc<dyn DataFrameTimeBounds>

sort_order: Option<Vec<String>>

The sort guarantee this partition’s rows will carry, per the caller’s data_sql’s ORDER BY (e.g. Some(["insert_time"]) for BlocksView). Recorded on Partition as-is.

unfiltered_count: Option<i64>

The unfiltered row count over the same join and range as source_count_query, read from that query’s unfiltered column when present – Some only for BlocksView, None for every other caller of this module. Self::write combines it with record_count to log how many rows the audience-mismatch predicate excluded; the check lives there rather than in make_batch_partition_spec so it only fires for partitions actually written, not every scheduled maintenance pass.

Trait Implementations

impl Debug for MetadataPartitionSpec

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

Formats the value using the given formatter.

impl PartitionSpec for MetadataPartitionSpec

fn is_empty(&self) -> bool

Returns true if the partition is empty.

fn get_source_data_hash(&self) -> Vec<u8>

Returns a hash of the source data.

fn write<'life0, 'async_trait>( &'life0 self, lake: Arc<DataLakeConnection>, logger: Arc<dyn Logger>, ) -> Pin<Box<dyn Future<Output = Result<(), Error>> + Send + 'async_trait>>

where 'life0: 'async_trait, MetadataPartitionSpec: 'async_trait,

Writes the partition to the data lake.