Trait PolicyEngine 

pub trait PolicyEngine: Send + Sync {
    // Required methods
    fn outbound_decision(&self, ctx: OutboundCtx<'_>) -> PolicyDecision;
    fn inbound_decision(&self, ctx: InboundCtx<'_>) -> PolicyDecision;
    fn accept_peer(&self, caps: &PeerCapabilities) -> bool;
}

Policy engine: decides the protection level for a concrete (peer, topic, interface) triple.

Safety classification

The trait is Send + Sync so it can be used via Arc<dyn PolicyEngine> in a multi-thread runtime. This triggers zerodds-lint: allow no_dyn_in_safe (documented in 08_heterogeneous_security.md §7).

Default contract

• Implementations must be deterministic: same context inputs → same decision. No randomness, no time-dependent branches (otherwise replay attacks are possible).
• accept_peer may return false if the peer does not meet the minimal requirements (e.g. a missing auth_plugin_class for a domain with allow_unauthenticated_participants=false).
• outbound_decision/inbound_decision must not block — they run in the hot path.

Required Methods

fn outbound_decision(&self, ctx: OutboundCtx<'_>) -> PolicyDecision

Outbound path: which protection level should the wire packet have?

fn inbound_decision(&self, ctx: InboundCtx<'_>) -> PolicyDecision

Inbound path: accept / drop / decrypt the packet?

fn accept_peer(&self, caps: &PeerCapabilities) -> bool

SEDP admission: is this peer (according to its capabilities) fundamentally acceptable for a match?

Dyn Compatibility

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors

impl PolicyEngine for GovernancePolicyEngine