fortran

A Fortran runner with automatic dependency resolution, intelligent caching, and frontend analysis support.

Our experimental lowercase fortran dialect pushes beyond all alternative scientific computing languages to combine Fortran's performance with modern expressiveness!

Quick Start

# Run any Fortran program instantly
fortrun hello.f90

# *lowercase fortran* .lf files (no boilerplate needed!)
fortrun script.lf

# Notebook mode - now available via fortbook package
# Install: https://github.com/lazy-fortran/fortbook

# Verbose mode for debugging
fortrun -v myprogram.f90

# Cache management
fortrun --clear-cache               # Clear all cached files
fortrun --clear-cache example.f90   # Clear cache and run
fortrun --cache-info                # Show cache statistics

# Compilation - now available via ffc package
# Install: https://github.com/lazy-fortran/ffc

lowercase fortran Showcase

Write fortran code with zero boilerplate - just the logic you need:

x = 5.0
y = 3.0
z = sqrt(x**2 + y**2)
print *, "Distance:", z

result = distance(3.0, 4.0)
print *, "Function result:", result

real function distance(a, b)
  real :: a, b
  distance = sqrt(a**2 + b**2)
end function distance

Runs as: fortrun calculate.lf

Automatically transforms to: - ✅ Wrapped in program statement - ✅ Modern implicit typing via automatic type inference - ✅ Double precision defaults (real(8)) - ✅ Type declarations automatically generated - ✅ contains section for functions - ✅ Function parameters default to intent(in) for safety

Features

🚀 Zero Configuration - No Makefiles, build scripts, or project setup - Automatic dependency detection and resolution - Smart caching with 2-4x performance improvements

🎯 Two Execution Modes
- Standard Fortran (.f90 files): Modern Fortran with no opinionated changes - lowercase fortran (.lf files): Compiler frontend with automatic type inference, no boilerplate, real(8) defaults

📦 Smart Dependencies - Local modules: Auto-includes Fortran files from same directory - Package registry: Resolves external modules to git repositories
- FPM integration: Leverages existing Fortran ecosystem

🚀 Advanced Features - Type Inference: Automatic variable declarations in .lf files (via fortfront) - Incremental Compilation: Only rebuilds changed files - Cache Management: CLI commands to clear cache and view statistics

📦 Extended Ecosystem - fortfront: Core analysis frontend (lexer, parser, semantic analysis) - ffc: Fortran Fortran Compiler - MLIR backend for compilation via HLFIR/LLVM - fnb: Fortran notebook interface - jupytext-style interactive computing with markdown/PDF output

Examples

Installation

git clone https://github.com/krystophny/fortran
cd fortran
./install.sh

Documentation

• 📋 ROADMAP.md - Development phases and future plans
• 📝 TODO.md - Current development tasks and progress
• 🏗️ CLAUDE.md - Technical implementation details

Usage

# Two execution modes
fortrun program.f90           # Standard Fortran mode
fortrun script.lf              # lowercase fortran mode (compiler frontend)

# Notebook mode - moved to fnb package
# fnb analysis.lf # Jupytext-style interactive analysis

# Options
fortrun -v file.f90           # Verbose output
fortrun --cache-dir DIR file  # Custom cache directory
fortrun --flag "-O3" file.f90 # Pass custom flags to FPM compiler

# Help
fortrun --help                # Show all options

Cache Management

The fortrun tool uses an intelligent caching system to speed up repeated builds. When developing or testing, you may need to manage the cache:

# View cache information
fortrun --cache-info
# Output:
# Fortran Cache Information:
#   Cache directory: /home/user/.cache/fortrun
#   Number of files: 487 files
#   Total size: 3.9M

# Clear entire cache
fortrun --clear-cache

# Clear cache and run a file (useful for testing)
fortrun --clear-cache example.f90

# Use custom cache directory
fortrun --cache-dir /tmp/fortrun-cache example.f90

Important for Development: Always clear the cache when testing new frontend features or compiler flags to ensure you're not using outdated cached files.

Project Status

Current: Production AST Frontend Complete ✅ - ✅ Complete 4-phase compiler architecture (lexer → parser → semantic → codegen) - ✅ Hindley-Milner type inference system with Algorithm W - ✅ Smart caching system (2-4x speedup) - ✅ lowercase fortran .lf syntax with automatic type inference - ✅ Notebook mode with basic output capture (figure capture WIP) - ✅ Comprehensive test suite (30+ frontend tests) - ✅ Parallel test runner for faster development

Next: FPM Registry integration and advanced language features

"Fortran is the Python of scientific computing - it just doesn't know it yet."