Skip to content

Latest commit

 

History

History
114 lines (91 loc) · 3.07 KB

CLAUDE.md

File metadata and controls

114 lines (91 loc) · 3.07 KB

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

NanoSTL is a portable, lightweight subset of C++ STL designed for constrained environments including embedded systems, CUDA devices, and JIT compilers. It's a header-only library with selective implementation requirements.

Build System and Commands

Testing

# Build and run unit tests
cd test
make
./tester

# Alternative CMake build
cd test
cmake .
make
./test_nanostl

Single Header Generation

# Generate consolidated single header file
python scripts/generateSingleHeader.py
# Output: single_include/nanostl.h

CMake Build (Static Library)

mkdir build && cd build
cmake ..
make

Architecture

Core Structure

  • include/: Main header files - each STL component is in nano*.h files
  • src/: Implementation files for components requiring NANOSTL_IMPLEMENTATION
  • single_include/: Generated consolidated header file
  • compat/: Standard STL compatibility layer

Key Components

  • nanovector.h: STL vector implementation
  • nanostring.h: String class with ryu-based float conversion
  • nanoalgorithm.h: STL algorithms (sort, find, etc.)
  • nanomath.h: Math functions (non-IEEE754 compliant approximations)
  • nanomap.h: Associative container
  • nanolimits.h: Numeric limits

Header Organization

Each header follows this pattern for single-header generation:

#ifndef NANOSTL_*_H_
#define NANOSTL_*_H_
// ... implementation ...
#endif // NANOSTL_*_H_

Implementation Pattern

For functions requiring implementation, use:

#define NANOSTL_IMPLEMENTATION
#include <nanostl.h>

Platform Support

Supported Platforms

  • 64-bit architectures only
  • Big and little endian
  • CUDA device functions
  • Windows, Linux, macOS
  • C++11 compilers (gcc 4.8+, clang 3.4+)

Compiler Macros

  • NANOSTL_BIG_ENDIAN: Set for big endian systems
  • NANOSTL_NO_IO: Disable I/O operations
  • NANOSTL_USE_EXCEPTION: Enable exception support
  • NANOSTL_NO_THREAD: Disable threading features
  • NANOSTL_PSTL: Enable parallel STL (requires C++17)
  • NANOSTL_IMPLEMENTATION: Include function implementations

Important Limitations

Thread Safety

  • Not thread-safe by default - applications must handle synchronization
  • Use external mutex/locks for concurrent access to containers

Memory Management

  • Returns NULL on allocation failure (no bad_alloc exception)
  • No automatic exception handling unless NANOSTL_USE_EXCEPTION is defined

Standards Compliance

  • Not IEEE-754 compliant for math functions
  • Limited STL compatibility - check feature matrix in README
  • No RTTI or exception support by default

Development Notes

Code Style

  • Use CUDA qualifiers for device compatibility
  • Follow existing naming conventions (nano* prefixes)
  • Implement both header and source file versions where needed
  • Maintain compatibility with C++11

Testing

  • Unit tests require C++11 compiler
  • Test both header-only and compiled modes
  • Include CUDA platform testing where applicable