statecraft.
Catalog/rustev

043-live-re-execution-at-scale

Live re-execution at dataset scale with spend caps (amends 026 and 006)

Lifecycle approvedImplementation completeVerification unknownQualification unknown

An amendment of approved spec 026 (live re-execution) and of spec 006 (CLI), for #76 child E3 (#74). A new crate `rustev-live` runs a dataset, or the inputs (never the outputs) of retained replay bundles, through one or more candidate plans live, through spec 026's `reexecute_live`, under an explicit `--live` opt-in, a per-run spend cap and the persistent spend journal of spec 012, generalized to any backend with a cost model and moved into `rustev-live`. A runtime spend gate durably reserves every attempt's full worst-case exposure before any byte is sent and refuses an attempt that would pass the run cap or the journal's cap (R-29: USD 10 in total for all testing across providers since A-75 raised it from USD 5). Acquisitions are declared live or captured per source; a probe checks reachability and served identity of every candidate binding before any dataset case is sent; scheduling is bounded per binding; runs resume from a checkpoint without re-sending a completed case or counting a charge twice. Every result is new evidence marked `live_reexecution` with a link to its source case or record, and per-run outputs are spec 029 repeat outputs with source `live`, so E1 and E2 consume them unchanged. Approved (A-40); approval authorizes no live run, provider call or spend by itself, raises no cap, and makes no quality claim.

Read the full pinned specification ↗

Relationships

Claimed units

  • establishes
    {
  "kind": "directory",
  "path": "crates/rustev-live/"
}
  • extends
    {
  "nature": "amending",
  "spec": "002-decision-contract-and-pure-core",
  "unit": {
    "kind": "directory",
    "path": "crates/rustev-contract/"
  }
}
  • extends
    {
  "nature": "amending",
  "spec": "003-runtime-execution-and-evidence",
  "unit": {
    "kind": "directory",
    "path": "crates/rustev-runtime/"
  }
}
  • extends
    {
  "nature": "amending",
  "spec": "004-evaluation-and-replay",
  "unit": {
    "kind": "directory",
    "path": "crates/rustev-eval/"
  }
}
  • extends
    {
  "nature": "amending",
  "spec": "006-cli-surface",
  "unit": {
    "kind": "directory",
    "path": "crates/rustev-cli/"
  }
}
  • extends
    {
  "nature": "additive",
  "spec": "012-jev-integration",
  "unit": {
    "kind": "directory",
    "path": "integrations/rustev-jev/"
  }
}
  • extends
    {
  "nature": "additive",
  "spec": "030-bedrock-backend",
  "unit": {
    "kind": "directory",
    "path": "integrations/rustev-bedrock/"
  }
}
  • extends
    {
  "nature": "additive",
  "spec": "031-public-api-and-release-family",
  "unit": {
    "kind": "directory",
    "path": "tools/rustev-release/"
  }
}
  • extends
    {
  "nature": "additive",
  "spec": "001-boundaries-and-authority",
  "unit": {
    "kind": "file",
    "path": "Cargo.toml"
  }
}
  • extends
    {
  "nature": "additive",
  "spec": "001-boundaries-and-authority",
  "unit": {
    "kind": "file",
    "path": "Cargo.lock"
  }
}
  • extends
    {
  "nature": "additive",
  "spec": "001-boundaries-and-authority",
  "unit": {
    "kind": "file",
    "path": "Makefile"
  }
}
  • extends
    {
  "nature": "additive",
  "spec": "001-boundaries-and-authority",
  "unit": {
    "kind": "directory",
    "path": "scripts/acceptance/"
  }
}

Source headings

  1. 043: Live re-execution at dataset scale with spend caps (amends 026 and 006)
  2. 1. Purpose
  3. 2. Territory
  4. 2.1 Binding principles of #76
  5. 3. Behavior
  6. 3.1 Live evaluation entry points
  7. 3.2 Live configuration
  8. 3.3 Inputs and snapshots
  9. 3.4 Live or captured acquisitions
  10. 3.5 The persistent spend journal
  11. 3.6 Spend gate and refusal before sending
  12. 3.7 Probe
  13. 3.8 Scheduling: concurrency, rate limits and deadlines
  14. 3.9 Evidence of live re-execution
  15. 3.10 Checkpoints and resumption
  16. 3.11 Reports into E1 and E2
  17. 3.12 Authorization in the CLI
  18. 3.13 CLI surface (amends spec 006)
  19. 3.14 Exit codes
  20. 3.15 Compatibility and identities
  21. 4. Out of scope
  22. 5. Observable negative cases
  23. Acceptance
  24. Verification
  25. Implementation record
  26. Decisions at approval

Frozen anchors

No frozen anchors declared.