Skip to content

Latest commit

 

History

History
45 lines (39 loc) · 4.32 KB

V1_FEATURE_COMPARISON.md

File metadata and controls

45 lines (39 loc) · 4.32 KB

Version 1 feature review

Version 1.0.0 was published on 2026-11-10 after the blockers below were resolved. See release evidence for the signed bundle and public-consumer checks.

Reviewed 2026-11-02 against current upstream main commits (read without changing sibling working trees): Ruby 288645b, Python 62d620a, TypeScript 961ceed. Java starts from the reviewed #85 head d2ee1a1; candidate changes are recorded in the linked issues. This compares documented capabilities and implementation entry points, not identical semantics or independent certification of the sibling libraries.

Capability Ruby / Python / TypeScript Java v1 candidate Decision
Dependency direction, cycles, naming Present in all three Present; ordinary-member cycle bug fixed in #85 Required
Empty-selection protection, exclusions, rationale Present in all three Present; strict imports retain errors Required
Layers, slices, PlantUML component rules Present in all three Present; bounded diagram subset Required
Graph reports All have DOT, Mermaid, D2, CSV, JSON, HTML Same six graph formats; bounded/escaped deterministic output Present
Test integration and custom rules Native frameworks and custom predicates JUnit assertions/engine and ArchitectureRules.define Present
Counts and dependency/distance metrics Present; formulas and source models vary Declaration/source-line counts, coupling, instability, abstractness, distance and cumulative metrics Present
Cohesion metrics Ruby/Python expose eight named variants; TS documents LCOM96a/b CK LCOM1, LCOM4, Henderson–Sellers; explicit availability policies Three supported formulas suffice; additional variants later
Metric thresholds and extension Siblings offer broader fluent comparisons/custom metric functions Inclusive at-most/at-least thresholds over fixed metric IDs; custom rules can evaluate other calculations Broader metric API later
Offline metrics HTML Present in all three Graph HTML and structured rule reports; no dedicated metrics dashboard Later enhancement
Logging and readable failure output Present in all three Opt-in inspection and console formatting since the unreleased diagnostics change Present
Language-specific model Ruby require/constants; Python imports/type checking; TS imports/Nx Bytecode members/calls, annotations, generics, records, sealed types, JPMS, lambdas Preserve Java semantics rather than copy source-language features
Reviewed baselines and advanced Java rules Not a uniform family feature Baselines/suppressions, reachability, member access, public interface rules, presets Already beyond the common feature floor
CLI/build use Ruby upstream lists standalone CLI as roadmap; Python config; TS framework API Properties/JUnit and CLI; add cycle kind in #89; Maven/Gradle invocation bridges No standalone build plugins required for v1

Release blockers and work order

  1. #85: preserve ordinary cycle evidence; merged after 372 local tests and successful upstream CI/CodeQL.
  2. #86: iterative cycle traversal; three 10,000-node regressions first demonstrated stack overflow.
  3. #89: declarative type/package cycle rules, exercised on compiled Java through all result formats.
  4. #88: explicit compatibility scope and reviewed API baseline before freezing 1.0.
  5. #87: metadata and final candidate gates, including upstream Linux/Windows CI and the protected publishing process.

Java 25 remains the supported runtime. A lower runtime floor would require a separately justified backend project. Full sibling metric parity, metrics HTML, and independently published Maven/Gradle plugins are useful follow-ups, not reasons to delay a correct, tested and clearly scoped v1.