Skip to content

v0.24.0

Choose a tag to compare

@HardMax71 HardMax71 released this 20 Aug 22:48
· 211 commits to main since this release
Immutable release. Only release title and notes can be modified.

Changes

  • Derive closed vocabulary spellings
  • Narrow repository open errors
  • Split human projections
  • Streamline invocation parsing
  • Derive wire control taxonomies
  • Split wire report subsystems
  • Compact immutable JSON values
  • Borrow CLI report projections
  • Use Result-based external command inputs
  • Split pack entry decoder
  • Split pack index decoder
  • Reuse loose object inflater state
  • Compact Git pack index lookups
  • Reject duplicate Git tree names
  • Split report analysis projection
  • Split report identity projection
  • Split anchor identity engine
  • Split correlation engine
  • Split scanner evaluation subsystems
  • Split scanner policy subsystems
  • Consolidate report taxonomies
  • Split resolution subsystems
  • Split scan pipeline orchestration
  • Split evaluation by responsibility
  • Derive report enum spellings
  • Split report construction by responsibility
  • Consume comparisons during report construction
  • Reduce scan allocation churn
  • Keep rendered observation identities coherent
  • Stream observation identity hashes
  • Resolve native paths in one buffer
  • Avoid transient resolver allocations
  • Count report payload once
  • Correct case-neighbor measurement scope
  • Make case-neighbor lookup allocation-free
  • Aggregate report summaries in one pass
  • Remove unreachable correlation duplicate check
  • Index correlation component roots
  • Split Markdown extraction subsystems
  • Split evaluation request engine
  • Split external plan engine
  • Split external assessment engine
  • Split control document models
  • Encode digest wire form once

Changes

  • Expose validated provider-lane object identities directly and build their execution constraint once in shared fixtures
  • Bind queued provider configuration directly into one shared repository-lane setup, removing parallel runtime assembly and automatic error conversions
  • Share forge verification facts and operation deadlines directly across provider transports, removing their parallel result enums and deadline implementations
  • Derive closed wire and scanner vocabulary parsing and projection with Strum, keep control trust typed through evaluation, and collapse adapter properties into one immutable metadata record
  • Separate repository-opening failures from post-open Git access defects and project repository unavailability only at process boundaries
  • Split the wire report subsystems and consolidate their taxonomies behind Strum spellings and immutable FindingKind::metadata(), replacing the public as_str, fixed_phase, scope, evidence_class, and invariant_class projection helpers
  • Split scanner policy acquisition, effects, and floor enforcement into focused modules, and remove the unused public floor_protected batch helper
  • Split scanner evaluation models and execution into focused modules, consolidate resolution-to-finding classification, and replace Attribution::as_str with Strum's AsRef<str> projection
  • Split wire control taxonomies from document decoding, replace their public as_str helpers and Profile::decode with Strum projections and parsing, and share one typed enum decoder internally
  • Split command-line argument collection from semantic classification, derive the closed verb and output-format parsers, and use canonical analysis-error identities and descriptions directly instead of mapping a duplicate invocation enum
  • Split correlation data, component-graph construction, and comparison derivation into focused modules, derive correlation-reason spellings, and avoid temporary removed/added document vectors while detecting exact renames
  • Separate the declarative renderer anchor contract from identity construction and duplicate resolution while preserving the public anchor API
  • Separate report evaluation identity and control provenance projection from observation and finding projection, share repository-identity serialization, and preserve the public candidate-identity digest API
  • Separate report observation, finding, and feedback projection from its public data model, project debt and waiver applications together with shared common rows, and inline the one-use location projection