Skip to content

CHANGELOG.md

Latest commit

 

History

History
98 lines (92 loc) · 7.68 KB

CHANGELOG.md

File metadata and controls

98 lines (92 loc) · 7.68 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and the project intends to use Semantic Versioning once executable releases begin.

Unreleased

Added

  • Added a verified, non-cloneable target-instance lifecycle that binds adapter, target-build, owner-attempt, and bounded instance identity; admits only typed prepare, execute, reset, cleanup, and discard transitions; prevents execution reuse before a confident reset; and now surrounds the real Linux CLI adapter with correlated prepared, executing, and cleaned/discarded events.
  • Completed the §§71–80 control-plane slice: a verified bounded scheduler; portable storage interfaces and conservative collection; schema-version-4 domain, replay, scenario, and persistence records; the full initial CLI grammar; fact-separated portable reports; five-seed replay evidence; structured tracing; known-defect and untrusted-boundary corpora; and explicit every-commit, nightly, and weekly CI tiers. The work is decomposed in a checked acceptance ledger and recorded by ADR-0005.
  • Split the normative design specification into concern-oriented authoritative slices behind a stable index, with documentation checks that reject missing, duplicated, reordered, or incorrectly owned numbered sections.
  • Added the first production crucible binary and crucible init [path], creating the documented workspace layout plus an application-identified, versioned, integrity-checked SQLite database; initialization is idempotent, refuses incompatible or symlinked state, and is covered through compiled-binary process fixtures.
  • Added monotonic workspace schema migration from v1 to v2 plus crucible artifact import and crucible artifact verify: verified canonical SHA-256 addressing feeds atomic no-clobber object publication, deduplication, source provenance, transactional database references, bounded reads, and post-publication corruption detection.
  • Complete draft implementation specification for the Crucible correctness harness.
  • Verus-first implementation saturation and trusted-boundary accounting requirements.
  • Project-owned, Verus-authored Crucible YAML design and proof obligations.
  • Evidence/provenance and multi-participant scenario architecture.
  • Public governance, security, support, contribution, roadmap, and community-health policies.
  • Documentation, dependency-review, and OpenSSF Scorecard automation.
  • Pinned Verus workspace, verified typed identifiers, strict trusted-boundary policy, and code CI.
  • Project-owned verified SHA-256, complete canonical digest decoding, algorithm-labeled artifact-ID parsing and dispatch, checksum-pinned NIST CAVP vectors, and content/size integrity checks.
  • Verified structurally validated append-only evidence/provenance publication, typed identity conflicts and missing endpoints, constructor-only portable UTC timestamps, exhaustive stable-tag equality, retry-safe borrowed payloads, normative relation direction, atomic multi-input derived evidence, explicit no-configuration identity, and versioned envelopes.
  • Project-owned Verus Crucible YAML profile-1 UTF-8 decoding with exact original-byte spans, explicit BOM policy, CR/CRLF normalization, typed malformed-input diagnostics, absolute and caller-lowered resource caps, private invariant-bearing constructors, a total pure success-or-error specification, and exact executable correspondence proofs.
  • Verified profile-1 lexical atomization with exhaustive YAML-indicator classification, one-to-one decoded-scalar and source-span preservation, private constructors, bounded all-or-error output, exhaustive Unicode tests, and total executable-to-pure correspondence.
  • Verified profile-1 line-layout analysis with exact atom and byte ranges, space-only indentation measurement, lossless leading-tab preservation for contextual scalar/separation decisions, deterministic resource-limit diagnostics, iterative progress, maximum-boundary fixtures, and total executable-to-pure correspondence.
  • Verified profile-1 structural-candidate partitioning with canonical-layout authentication, exact lossless atom/byte coverage, directives and document markers, separation/comment/flow candidates, caller-lowered all-or-error bounds, typed mismatch and first-excluded diagnostics, iterative progress, a total pure model, and exact executable correspondence for success and failure.
  • Verified profile-1 context-sensitive single- and double-quoted scalar boundaries with canonical upstream authentication, provisional plain/block-region tracking, complete YAML 1.2 escape validation, exact source ranges, caller-lowered all-or-error caps, public semantic/range proofs, and total executable correspondence for success and every typed failure.
  • Verified profile-1 context-sensitive plain-scalar boundaries with retained multiline presentation, flow-sensitive indicator and mapping-colon handling, node-property and block-region exclusion, contextual leading-tab diagnostics, exact source ranges, caller-lowered all-or-error caps, public semantic/range proofs, and total executable correspondence. Adversarial fixtures cover verbatim tag punctuation, coalesced flow colons, tab-only prefixes, block-header comments, and malformed ?, :, and - scalar starts.
  • Verified profile-1 literal and folded block-scalar formation with complete YAML 1.2 header, contextual compact-collection indentation, folding, and strip/clip/keep chomping behavior; contextual tabs; exact raw ranges; per-code-point direct/folded source provenance; independent scalar, presentation, scalar-content, and total-content caps; distinct typed upstream evidence diagnostics; a total pure model; exact executable correspondence for every success and failure; and a general proof that every authenticated nonempty success has exact rendered content and ordered non-overlapping atom/byte ranges.
  • Verified profile-1 completed-token formation with canonical authentication of every preceding lexer witness; a lossless adjacent atom/byte partition; retained trivia and exact directive, property, alias, tag, and scalar identities; typed contextual indicators and bounded flow-stack validation; exact tag-character and BOM-context diagnostics; caller-lowered absolute limits; a total pure result model; exact executable correspondence; and public partition, formation, balance, anti-laundering, scalar-identity, parts, trivia-maximality, and limit proof surfaces.
  • Verified alias-transparent canonical YAML DAG lowering after duplicate rejection and merge expansion, with stable source-node identity, normalized roots and collection edges, retained scalar/collection tags and merge provenance, graph sharing, four independent resource caps, a total pure result model, exact executable correspondence, and public anti-forgery contracts.
  • Added verified compilation for the versioned typed-field schema graph consumed by configuration lowering, including every scalar and custom-tagged collection kind, nested sequence/mapping references, required-field metadata, stable field IDs, Unicode names, ambiguity rejection, caller-lowered resource caps, exact pure semantics, and public identity/uniqueness contracts.
  • Added verified canonical-YAML value binding against every typed schema kind, with exact Core and custom tag separation, distinct finite/nonfinite scalar variants, alias-transparent resolved-node identity, scalar/collection record authentication, source-anchored typed diagnostics, a total pure result model, and public deterministic-output anti-forgery contracts.