Enum Command 

pub enum Command {
    DefineSchema {
        name: String,
        columns: Vec<ColumnSpec>,
    },
    WriteSingleTs {
        shard: String,
        schema: Option<String>,
        ts: u64,
        count: u64,
        start: u64,
        row_bytes: Option<usize>,
    },
    WriteSpread {
        shard: String,
        schema: Option<String>,
        count: u64,
        n_ts: u64,
        start: u64,
        row_bytes: Option<usize>,
    },
    WriteRows {
        shard: String,
        schema: Option<String>,
        ts: u64,
        rows: Vec<Vec<String>>,
    },
    DefineIndex {
        source_id: u64,
        index_id: u64,
        shard: String,
        schema: Option<String>,
        key: Vec<usize>,
        as_of: u64,
        upper: u64,
    },
    Schedule {
        id: u64,
    },
    AllowCompaction {
        id: u64,
        frontier: u64,
    },
    AllowWrites {
        id: u64,
    },
    AwaitFrontier {
        id: u64,
        ts: u64,
        timeout_secs: Option<u64>,
        allow_timeout: bool,
    },
    Count {
        id: u64,
        ts: u64,
    },
    CreateDataflow {
        name: Option<String>,
        imports: Vec<ImportSpec>,
        builds: Vec<BuildSpec>,
        exports: Vec<ExportSpec>,
        as_of: u64,
        optimize: bool,
    },
    Peek {
        id: u64,
        schema: Option<String>,
        ts: u64,
    },
    AwaitSubscribe {
        id: u64,
        up_to: u64,
        timeout_secs: Option<u64>,
    },
    CreateInstance {
        expiration_offset: Option<String>,
        arrangement_dictionary_compression: bool,
    },
    UpdateConfiguration {
        updates: Vec<ConfigSetting>,
    },
    Reconnect,
    InitializationComplete,
}

A command read from the script stream.

Tagged on "cmd", snake_case, e.g. {"cmd":"write_single_ts","shard":"s1","ts":0,"rows":1000}.

Variants

DefineSchema

Declare a named relation schema for later schema references.

Fields

name: String

Schema name, referenced by schema fields on other commands.

columns: Vec<ColumnSpec>

Ordered column declarations.

WriteSingleTs

Write count synthetic rows to shard at a single timestamp ts.

Fields

shard: String

Shard alias; allocated on first use.

schema: Option<String>

Schema name; defaults to the built-in (bigint, text) sample schema.

ts: u64

The timestamp to write at.

count: u64

Number of synthetic rows to write.

start: u64

First synthetic row index, so successive batches can use disjoint id ranges (start..start + count) that never consolidate. Defaults to 0.

row_bytes: Option<usize>

Payload padding per row; defaults to DEFAULT_ROW_BYTES.

WriteSpread

Write count synthetic rows to shard, spread across n_ts timestamps in a single append.

Fields

shard: String

Shard alias; allocated on first use.

schema: Option<String>

Schema name; defaults to the built-in sample schema.

count: u64

Number of synthetic rows to write.

n_ts: u64

Number of distinct timestamps to spread the rows across.

start: u64

First synthetic row index (see Command::WriteSingleTs). Defaults to 0.

row_bytes: Option<usize>

Payload padding per row; defaults to DEFAULT_ROW_BYTES.

WriteRows

Write explicit rows to shard at a single timestamp ts. Each row is an array of JSON values matching the schema’s columns in order.

Fields

shard: String

Shard alias; allocated on first use.

schema: Option<String>

Schema name; defaults to the built-in sample schema.

ts: u64

The timestamp to write at.

rows: Vec<Vec<String>>

Rows as arrays of raw value tokens, typed against the schema by cell_from_token.

DefineIndex

Submit (without scheduling) an index dataflow over shard.

Fields

source_id: u64

The imported source’s global id.

index_id: u64

The exported index’s global id.

shard: String

Shard alias to import; must already exist.

schema: Option<String>

Schema name; defaults to the built-in sample schema. Must match what was written to shard.

key: Vec<usize>

Columns to arrange by.

as_of: u64

The dataflow’s as_of.

upper: u64

The shard’s exclusive write upper (see PersistSource::upper).

Schedule

Schedule a previously-submitted collection so it makes progress.

Fields

id: u64

The collection’s global id.

AllowCompaction

Advance an index’s read frontier (since) via AllowCompaction.

Fields

id: u64

The index’s global id.

frontier: u64

The new read frontier.

AllowWrites

Take a collection out of read-only mode via AllowWrites, letting its persist sink begin writing. Every dataflow starts read-only; indexes, subscribes, and peeks work regardless, but a materialized-view sink withholds all persist writes until this is sent for its sink id.

Fields

id: u64

The sink’s global id.

AwaitFrontier

Wait until id’s output frontier reaches ts, or fail after the timeout.

Fields

id: u64

The collection’s global id.

ts: u64

The target output-frontier timestamp.

timeout_secs: Option<u64>

Timeout in seconds; defaults to DEFAULT_TIMEOUT_SECS.

allow_timeout: bool

If true, a timeout is reported (status: timeout) without failing the run. Used by reproductions where not reaching the frontier is an expected outcome, not an assertion failure.

Count

Count id’s rows at ts and emit the count.

Sugar over a Reduce: builds an ephemeral dataflow that index-imports id, computes count(*) over it, and peeks the single-row result — so the count runs through a real reduce operator rather than being tallied in the driver. id must have been registered by a prior define_index (or define export). The golden output is the count; the script’s ---- block asserts it.

Fields

id: u64

The index’s global id.

ts: u64

The timestamp to count at.

CreateDataflow

Submit (without scheduling) a dataflow built from generic MIR — the abstraction behind index / materialized-view / subscribe / copy-to.

A projection of DataflowBuilder: import sources and/or existing indexes, build MIR objects (each a pretty-form MIR spec; see BuildSpec), and export over them. Exports are index, materialized-view, or subscribe (see ExportSpec); copy-to is not implemented. Exported indexes are registered for later import or count assertion; subscribe sinks register a response buffer for await-subscribe. define-index is sugar over this. The optional optimize flag runs the MIR optimizer before lowering (needed for joins).

Fields

name: Option<String>

Debug name for the dataflow; defaults to headless-create-dataflow.

imports: Vec<ImportSpec>

Collections to import (persist sources and/or existing indexes).

builds: Vec<BuildSpec>

MIR objects to compute, each bound to an id.

exports: Vec<ExportSpec>

Exports over imported or built ids.

as_of: u64

The dataflow’s as_of.

optimize: bool

Run the MIR optimizer before lowering (needed for e.g. joins). Off by default, so the caller’s MIR is lowered faithfully.

Peek

Peek id at ts and emit the returned rows (sorted, one per line). The generic output assertion: the script’s ---- block holds the expected rows.

Fields

id: u64

The index’s global id.

schema: Option<String>

Schema name describing the peek’s output; defaults to the sample schema.

ts: u64

The timestamp to peek at.

AwaitSubscribe

Wait for subscribe sink id’s upper to reach up_to, then emit its accumulated updates as <ts> <diff> <datums> lines (consolidated, sorted). The output assertion for a subscribe sink.

Fields

id: u64

The subscribe sink’s global id.

up_to: u64

The exclusive upper to wait for (typically the sink’s up_to).

timeout_secs: Option<u64>

Timeout in seconds; defaults to DEFAULT_TIMEOUT_SECS.

CreateInstance

Send CreateInstance, opening the compute instance (and the reconciliation window). The settable InstanceConfig knobs default to the values a plain create-instance supplies; the peek-stash location is always the host’s, and the peek-response stash is force-disabled (see Driver::create_instance).

Fields

expiration_offset: Option<String>

Replica expiration offset (a duration like 30s); none if absent.

arrangement_dictionary_compression: bool

Whether arrangements use dictionary compression.

UpdateConfiguration

Send UpdateConfiguration with a table of dyncfg updates (name type value rows). Generic over any configuration; the peek-response stash is not settable here (it is force-disabled at instance creation).

Fields

updates: Vec<ConfigSetting>

The dyncfg updates to apply.

Reconnect

Drop the current connection and reconnect, sending only Hello. Re-issue create_instance, replay the dataflows the replica should keep, then initialization_complete to close the reconciliation window.

InitializationComplete

Send InitializationComplete, closing the reconciliation window.

Trait Implementations

impl Clone for Command

fn clone(&self) -> Command

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Command

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Command

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for Command

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for Command

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl Eq for Command

impl StructuralPartialEq for Command