Expand description
Conservative caller-owned aggregate admission for one protected root.
Charges survive every publication error and unwind. There is no refund or directory scan; recovery may construct a new ledger only from a freshly authenticated usage snapshot. This is cooperative in-process accounting, not a durable cross-process quota or a filesystem allocation limit.
Re-exports§
pub use recovery::RecoveryRootScope;pub use staging::StagingCharge;
Modules§
- recovery
- Zero-growth root provenance from the actual admitted quota owner. No marker/scan reconstructs usage, per-run membership or a budget token. Host holds current protected-namespace and aggregate-quota leases throughout.
- staging
- Staging admission shares the existing quota’s root, usage and attempt slots.
Structs§
- Custody
Quota - N bounds additional publication attempts tracked in this ledger. The host must hold its aggregate-quota lease for the ledger’s complete lifetime.
- Quota
Limits - Quota
Snapshot - Host snapshot must include committed, orphan and uncertain bundle charges. The opaque snapshot ID must bind the root and current usage under the host’s trusted freshness and lease policy. Counts exclude OS/runtime allocations.
Enums§
Traits§
Functions§
- publish_
bundle_ with_ quota - Admit aggregate encoded bundle bytes before any producer I/O. The encoded byte cap excludes filesystem overhead and head/lock files; host disk quotas must cover those. Authority still revalidates policy during publication.
- publish_
chain_ bundle_ with_ quota