Struct MultiEntityBytecode 

pub struct MultiEntityBytecode {
    pub entities: HashMap<String, EntityBytecode>,
    pub event_routing: HashMap<String, Vec<String>>,
    pub when_events: HashSet<String>,
    pub proto_router: ProtoRouter,
}

Fields

entities: HashMap<String, EntityBytecode>

event_routing: HashMap<String, Vec<String>>

when_events: HashSet<String>

proto_router: ProtoRouter

Implementations

impl MultiEntityBytecode

pub fn from_single<S>( entity_name: String, spec: TypedStreamSpec<S>, state_id: u32, ) -> Self

pub fn from_entities(entities_vec: Vec<(String, Box<dyn Any>, u32)>) -> Self

pub fn new() -> MultiEntityBytecodeBuilder

pub fn fingerprint(&self) -> String

Deterministic content hash of the compiled bytecode, used to invalidate state snapshots when the stack’s compiled logic changes.

Covers state ids, handler opcode streams, routing, and field metadata. Computed-field evaluators are opaque closures, so only their presence is hashed — their bodies are generated from the same stack source as the opcode streams, which change alongside them in practice.

Opcode streams are hashed via a canonical serde encoding (externally tagged variants, object keys sorted), NOT via Debug formatting, so the hash survives cosmetic churn that doesn’t alter the compiled program: rustc Debug formatting changes, struct field reordering, and inserting new OpCode variants. Renaming a variant or field DOES change the hash (names are the tags) — treat renames as snapshot-invalidating. Interpreter behavior changes that keep the encoding identical are invisible here; those must bump SNAPSHOT_FORMAT_VERSION instead.

Trait Implementations

impl Debug for MultiEntityBytecode

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

Formats the value using the given formatter.