Coordinate compiler implementation against executable test oracles

Why use this

1. A compiler research team must divide implementation work while keeping language behavior and test expectations consistent. Parallel changes can appear productive even when they overlap, invalidate shared assumptions or improve only a narrow set of examples.

2. This job could organize task ownership and test evidence around a clearly bounded research target. A reviewable comparison plan helps researchers distinguish supported language behavior, unresolved discrepancies and missing coverage before interpreting progress.

Why an agent / simpler alternative

Useful judgment

Coordinate dependencies, interpret test disagreements and identify coverage gaps without treating one oracle as unquestionable authority.

Simpler alternative

Use a conventional small-team workflow when implementation scope and ownership are already manageable.

Unsuitable when

The project would run untrusted code without isolation or claim compiler correctness from limited tests.

Proposed inputs

1. Name

   Language/project scope

   Owner

   Research lead

   Minimum scope

   Target features, exclusions and repository boundaries

   Quality check

   Define intended behavior and unsupported cases

2. Name

   Harness/test manifest

   Owner

   Authorized engineer

   Minimum scope

   Test identities, oracle versions and expected comparison rules

   Quality check

   Record environment assumptions and provenance

Access and setup — proposed or unknown

Proposed minimum

1. Review a fictional task board and test manifest; no code changes, containers, compiler execution or compute spend.

Compatibility unknowns

1. Current toolchain versions, isolation controls, oracle compatibility and runtime support are unknown.

External actions requiring approval

1. Repository writes, running generated programs, infrastructure changes, publishing or resource spending requires explicit authorization.

Expected deliverable and review checks

Proposed artifact

Coordination plan with task ownership, integration checkpoints, comparison categories and unresolved research questions.

Critical checks

1. Prevent overlapping edits; distinguish compile success from behavioral agreement; retain failed/unsupported cases and precise test scope.

Missing output evidence

This overlay adds no new output inspection, execution or validation; consult the referenced canonical source outcome limits.

Failure signals and proposed responses

1. Signal

   Two workers modify the same interface differently

   Why it matters

   Independent progress breaks integration

   Proposed response

   Assign interface ownership and reconcile before combining work.

2. Signal

   Oracle disagreement lacks environment context

   Why it matters

   A mismatch may be misclassified as compiler failure

   Proposed response

   Record versions and assumptions before interpretation.

Effort and cost — unknown

Setup work

1. Define feature boundaries and reproducible comparison conventions.

Ongoing review

1. Researchers review integration and authorized execution evidence.

Unknowns

1. Test completeness, language coverage, compute demand and coordination costs are unknown.

First check — proposed, unexecuted

Status

proposed_unexecuted

Synthetic fixture

Invent two overlapping implementation tasks and three test-manifest entries with a version mismatch and unsupported feature; review coordination only.

Acceptance checks

1. Resolve ownership; flag oracle context; claim no compiler tests or correctness validation.

Stop conditions

1. Stop if isolation, expected semantics or compute authority is unresolved.

Limits

1. A coordination plan cannot establish compiler correctness, language conformance or reliability beyond future inspected evidence.