Skip to content

Release v0.2.0 - Semantic Group Parallelization

Choose a tag to compare

View all tags
@mmcky mmcky released this 02 Oct 04:04
· 98 commits to main since this release
v0.2.0
9010cd4

Release v0.2.0 - Semantic Group Parallelization

🚀 Major Features

Semantic Group Parallelization

Performance Revolution: Style guide checking is now 2-3x faster and 25% cheaper through intelligent semantic grouping and parallel processing.

  • ⚡ 2-3x faster - Reviews complete in 6-8 seconds (vs 12-20 seconds)
  • 💰 25% cheaper - Reduced API costs (~$0.42 vs ~$0.56 per lecture)
  • 🎯 Better quality - Related rules checked together for improved accuracy
  • 🔄 Parallel processing - Up to 4 semantic groups reviewed simultaneously

Markdown-Based Style Guide Database

Migrated from YAML to Markdown for more natural LLM integration:

  • 📝 48 total rules organized in 8 semantic groups
  • 🎯 31 actionable rules (category='rule') - automatically fixed
  • 📚 13 advisory rules (category='style') - for future enhancement
  • 🔄 4 migration patterns (category='migrate') - transformation rules

8 Semantic Groups

Rules organized by content type for better context:

Group Actionable Rules Focus Area
WRITING 4 Prose structure, paragraphs
MATH 8 LaTeX formatting, equations
CODE 3 Code blocks, syntax
JAX 1 JAX-specific patterns
FIGURES 9 Figures, captions, alt text
REFERENCES 1 Citations
LINKS 1 URL formatting
ADMONITIONS 4 Note/warning blocks

✨ What's New

New Review Architecture

  • review_lecture_smart() - Semantic grouping orchestrator with parallel execution
  • Automatic group detection and parallel processing
  • Single-pass fix application across all violations
  • Resilient: Individual group failures don't break entire review

Enhanced Parser

  • Complete Markdown parser (parser_md.py) with 98% test coverage
  • Three-tier category system (rule/style/migrate)
  • Automatic semantic group extraction
  • Rich rule metadata with examples and guidance

Improved Code Quality

  • Simplified codebase: Removed ~109 lines of dead code
  • Single code path: No more chunking complexity
  • Better maintainability: Cleaner architecture
  • All tests passing: 10/10 tests (parser + semantic grouping)

🔧 Breaking Changes (Early Development)

Since this is early development, we've removed backwards compatibility for a cleaner codebase:

Removed:

  • review_in_chunks() method (replaced by review_lecture_smart())
  • format_rules_for_llm() function (no longer needed)
  • max_rules parameters (automatic semantic grouping)
  • --max-rules-per-request CLI argument
  • Old YAML parser and database

Migration: No action required for users - the action automatically uses the new system.

📊 Example Output

🤖 Starting AI-powered review using semantic grouping...
📊 Lecture: aiyagari.md
📋 Total actionable rules: 31

📦 Processing 8 semantic groups in parallel:
   • WRITING: 4 rules
   • MATH: 8 rules
   • CODE: 3 rules
   • JAX: 1 rule
   • FIGURES: 9 rules
   • REFERENCES: 1 rule
   • LINKS: 1 rule
   • ADMONITIONS: 4 rules

🚀 Running 4 parallel reviews...

  ✓ WRITING: 3 issues found
  ✓ MATH: 5 issues found
  ✓ CODE: 1 issue found
  ✓ JAX: No issues found
  ✓ FIGURES: 2 issues found
  ✓ REFERENCES: No issues found
  ✓ LINKS: No issues found
  ✓ ADMONITIONS: 1 issue found

📊 Total issues found across all groups: 12
  🔧 Applying 12 fixes programmatically...

✓ Review complete in 6.2 seconds
💰 Estimated cost: $0.42

🧪 Testing

  • ✅ All 7 parser tests passing (98% coverage)
  • ✅ All 3 semantic grouping tests passing
  • ✅ 10/10 total tests passing
  • ✅ No import errors
  • ✅ No broken references

📚 Files Changed

New:

  • style_checker/parser_md.py (+299 lines) - Markdown parser
  • tests/test_parser_md.py (+191 lines) - Parser tests
  • tests/test_semantic_grouping.py (+179 lines) - Integration tests
  • style-guide-database.md - New Markdown database

Modified:

  • style_checker/reviewer.py (+183, -76) - Semantic grouping implementation
  • style_checker/main.py (simplified by 36 lines)
  • CHANGELOG.md - Comprehensive v0.2.0 release notes

Removed:

  • style-guide.yaml (replaced by Markdown version)
  • style_checker/parser.py (replaced by parser_md.py)
  • tests/test_basic.py (replaced by test_parser_md.py)
  • ~109 lines of deprecated code

🎯 Impact

This release represents a major architectural improvement:

  • Faster: 2-3x speedup through parallel processing
  • Cheaper: 25% cost reduction through intelligent grouping
  • Better: Improved quality from semantic coherence
  • Cleaner: Single code path, no deprecated methods
  • Tested: Comprehensive test suite with 98% coverage

Perfect for early development and testing of the QuantEcon Style Guide Action!

📖 Full Details

See CHANGELOG.md for complete technical details, migration guide, and implementation notes.

Release Date: October 2, 2025
Status: Early Development / Testing
Recommended for: Development and testing environments
Floating Tag: Use v0.2 for latest v0.2.x updates

Assets 2
Loading