Struct MergeSpec 

pub struct MergeSpec { /* private fields */ }

A merge operation: several tables into one, with no user job.

What a sort does for order, a merge does for chunk layout — and in MergeMode::Sorted it is the cheap way to combine tables that are already sorted, because nothing has to be sorted again.

use ytsaurus_client::{MergeMode, MergeSpec};

let spec = MergeSpec::new(["//tmp/monday", "//tmp/tuesday"], "//tmp/week")
    .with_mode(MergeMode::Sorted)
    .with_merge_by(["host"]);

Implementations

impl MergeSpec

pub fn new<I>(inputs: I, output: impl Into<String>) -> Self

where I: IntoIterator, I::Item: Into<String>,

Merges inputs into output, unordered.

pub fn with_mode(self, mode: MergeMode) -> Self

Chooses how the inputs are combined.

pub fn with_merge_by<K>(self, columns: K) -> Self

where K: IntoIterator, K::Item: Into<String>,

The columns a MergeMode::Sorted merge merges by.

The output table comes back sorted by these. Optional: a sorted merge sent without it is accepted, and the cluster uses the sort columns the inputs already have. Naming them is how to merge by a prefix of that, and how to make the assumption visible where it is being made.

pub fn with_combine_chunks(self, combine: bool) -> Self

Asks the cluster to combine small chunks while it merges.

This is most of why a merge is worth running on one table: a table written in many small pieces reads faster afterwards.

pub fn with_force_transform(self, force: bool) -> Self

Runs the jobs even when the merge could be done by moving chunks.

A merge that has nothing to do normally just relinks chunks. Set this when the point is the rewrite — a change of compression codec or erasure coding, which only happens where rows are actually copied.

pub fn with_job_count(self, count: i64) -> Self

Asks for a particular number of jobs.

Takes precedence over data_size_per_job, which the cluster otherwise uses to decide.

pub fn with_raw(self, key: impl Into<String>, value: YsonValue) -> Self

Sets any spec field this builder does not model — data_size_per_job, schema_inference_mode and the rest.

pub fn to_yson(&self) -> YsonValue

Renders the spec.

Trait Implementations

impl Clone for MergeSpec

fn clone(&self) -> MergeSpec

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MergeSpec

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

Formats the value using the given formatter.