Skip to content

ROADMAP.md

Latest commit

 

History

History
112 lines (81 loc) · 4.15 KB

ROADMAP.md

File metadata and controls

112 lines (81 loc) · 4.15 KB

dialyzer_json Roadmap

Vision: AI-friendly JSON output for Dialyzer warnings, optimized for Claude Code and similar tools.

Completed work: See CHANGELOG.md for finished tasks.

🎯 Current Focus

Phase 5: Complete — All planned features implemented. Ready for hex.pm release.

✅ Recently Completed

Task Description Notes
Task 7 --group-by-file flag Groups by file path, array of {file, count, warnings}
Task 6 exact_compare → "code" 1-line fix
Task 5 Top-level metadata schema_version, versions, timestamp
Task 8 --compact JSONL output One warning per line
Task 9 --filter-type flag Repeatable, OR logic
Basic JSON output mix dialyzer.json outputs warnings as JSON Reuses dialyxir PLT
Quiet mode --quiet suppresses non-JSON output Clean piping to jq
Summary mode --summary-only for counts without details Quick health check
Group by warning --group-by-warning clusters similar warnings Pattern identification
Warning classification All 47 warning types classified DialyzerJson.FixHint module
fix_hint in output Each warning has fix_hint field Summary includes by_fix_hint
Tasks 3-4 Module/function extraction for contract/callback warnings 8 contract + 6 callback types

Phase 2: Warning Classification ✅

Added fix_hint field to help AI editors prioritize warnings.

  • Task 1: Classified all 47 dialyxir warning types into "spec", "code", "pattern" categories
  • Task 2: Added fix_hint to each warning and by_fix_hint to summary

Phase 3: Better Metadata Extraction ✅

Extended module and function extraction to cover all warning types.

  • Task 3: Module extracted from 8 contract warning types and 6 callback warning types ✅
  • Task 4: Function extracted from contract and callback warnings (guard_fail and pattern_match don't contain function info) ✅

Phase 4: Output Enhancements

Task 5: Add top-level metadata to JSON output ✅

Add metadata fields to the root JSON object for better tooling integration.

Fields to add:

  • schema_version: e.g. "1.0" (for future compatibility)
  • dialyzer_version: from :dialyzer app version
  • elixir_version: from System.version()
  • otp_version: from :erlang.system_info(:otp_release)
  • run_at: ISO8601 timestamp (optional, include by default)

Success criteria:

  • All version fields present in output
  • run_at is valid ISO8601 format
  • Existing warnings and summary structure unchanged

Task 6: Extend fix_hint mapping for exact_compare ✅

Map exact_compare warning type to "code" (currently falls through to unknown). These are actionable warnings about == vs === comparisons.

Success criteria:

  • exact_compare → "code" in FixHint module
  • Test covers the mapping

Task 7: Add --group-by-file flag ✅

Add --group-by-file flag to group warnings by file path instead of warning type.

Output format:

{
  "groups": [
    { "file": "lib/foo.ex", "count": 3, "warnings": [...] }
  ],
  "summary": { ... }
}

Success criteria:

  • --group-by-file groups warnings by file
  • Each group has file, count, and warnings fields
  • Default output (flat warnings) unchanged

Task 8: Add --compact flag for JSONL output ✅

Add --compact flag that outputs one JSON object per line (JSONL format) instead of a single JSON array. Useful for streaming large warning sets.

Success criteria:

  • --compact outputs one warning per line
  • Each line is valid JSON
  • Summary on final line in compact mode

Task 9: Add --filter-type flag ✅

Add --filter-type TYPE flag to output only warnings of a specific type. Can be repeated for multiple types.

Success criteria:

  • --filter-type no_return shows only no_return warnings
  • Multiple --filter-type flags combine with OR logic
  • Summary reflects filtered count

Phase 5: Documentation & Polish ✅

3 tasks complete. See CHANGELOG.md for details. Added: hex.pm metadata, AGENTS.md, comprehensive README examples.