Domain Flow Setup Guide

Quick Start

1. Install Prerequisites

Linux/macOS

• CMake 3.28+
• C++20 compiler (GCC 10+, Clang 12+)
• Ninja build system (optional but recommended)
• System dependencies (installed via package manager)

# Ubuntu/Debian
sudo apt install cmake ninja-build g++

# Fedora/RHEL
sudo dnf install cmake ninja-build gcc-c++

# macOS (with Homebrew)
brew install cmake ninja

Windows

• CMake 3.28+
• Visual Studio 2022 with C++ development tools OR Ninja
• vcpkg package manager

# Install vcpkg
git clone https://github.com/Microsoft/vcpkg.git
cd vcpkg
.\bootstrap-vcpkg.bat

# Set environment variable (PowerShell)
$env:VCPKG_ROOT = "C:\dev\vcpkg"
# Or set permanently via System Properties > Environment Variables

Optional (for MLIR tools, all platforms)

• LLVM/MLIR 17+ built from source

2. Configure CMake User Presets

Choose the template for your platform:

Linux (uses native system dependencies, no vcpkg)

cp CMakeUserPresets.json.Linux.template CMakeUserPresets.json

Edit CMakeUserPresets.json to set your LLVM path:

{
  "configurePresets": [
    {
      "name": "user-ninja-release",
      "inherits": "Ninja-Release",
      "environment": {
        "LLVM_PROJECT_ROOT": "/home/yourname/dev/clones/llvm-project"
      }
    }
  ]
}

Windows (uses vcpkg for dependencies)

cp CMakeUserPresets.json.Windows.template CMakeUserPresets.json

Edit CMakeUserPresets.json to set your local paths:

{
  "configurePresets": [
    {
      "name": "user-vs17-release",
      "inherits": "VS17-Release",
      "environment": {
        "VCPKG_ROOT": "C:/dev/vcpkg",
        "LLVM_PROJECT_ROOT": "C:/dev/clones/llvm-project"
      },
      "cacheVariables": {
        "CMAKE_TOOLCHAIN_FILE": {
          "type": "FILEPATH",
          "value": "$env{VCPKG_ROOT}/scripts/buildsystems/vcpkg.cmake"
        }
      }
    }
  ]
}

Note: CMakeUserPresets.json is gitignored and won’t be committed. The platform-specific templates are checked in to preserve configuration knowledge.

3. Configure and Build

Linux - Basic Build (no MLIR)

# Configure (uses native system packages)
cmake --preset user-ninja-release

# Build
cmake --build build/Ninja-Release

# Run tests (if enabled)
ctest --test-dir build/Ninja-Release

Windows - Basic Build (with vcpkg)

# Configure (uses vcpkg for dependencies)
cmake --preset user-vs17-release

# Build
cmake --build build_msvc/VS17-Release

# Or use Ninja on Windows
cmake --preset user-ninja-release
cmake --build build/Ninja-Release

All Platforms - With MLIR Tools

First, build LLVM/MLIR (Linux/macOS):

git clone https://github.com/llvm/llvm-project.git
cd llvm-project
mkdir build && cd build

cmake -G Ninja ../llvm \
  -DLLVM_ENABLE_PROJECTS="mlir" \
  -DCMAKE_BUILD_TYPE=Release \
  -DLLVM_TARGETS_TO_BUILD="X86" \
  -DLLVM_ENABLE_ASSERTIONS=ON

ninja

Windows (use build_msvc directory):

cd llvm-project
mkdir build_msvc
cd build_msvc

cmake -G "Visual Studio 17 2022" ..\llvm `
  -DLLVM_ENABLE_PROJECTS="mlir" `
  -DCMAKE_BUILD_TYPE=Release `
  -DLLVM_TARGETS_TO_BUILD="X86"

cmake --build . --config Release

Then configure Domain Flow with MLIR:

# Ensure CMakeUserPresets.json has:
# "LLVM_PROJECT_ROOT": "/path/to/llvm-project"

# Configure with MLIR tools enabled
cmake --preset user-ninja-release -DDOMAINFLOW_MLIR_TOOLS=ON

# Build
cmake --build build/Ninja-Release

Alternative: Environment Variables

Instead of using CMakeUserPresets.json, you can set environment variables:

# Linux/macOS
export VCPKG_ROOT=/path/to/vcpkg
export LLVM_PROJECT_ROOT=/path/to/llvm-project

# Then use the base presets
cmake --preset Ninja-Release

# Windows (PowerShell)
$env:VCPKG_ROOT = "C:\path\to\vcpkg"
$env:LLVM_PROJECT_ROOT = "C:\path\to\llvm-project"

cmake --preset Ninja-Release

Build Options

Control features with CMake options:

cmake --preset user-ninja-release \
  -DDOMAINFLOW_BUILD_TESTING=ON \
  -DDOMAINFLOW_TOOLS=ON \
  -DDOMAINFLOW_DSE=ON \
  -DDOMAINFLOW_MLIR_TOOLS=OFF

Available Options

Option	Default	Description
DOMAINFLOW_BUILD_TESTING	BUILD_TESTING (root) / OFF (subproject)	Build and register tests
DOMAINFLOW_TOOLS	OFF	Build dfg/rdg tools
DOMAINFLOW_POLYHEDRAL	ON	Build polyhedral tools
DOMAINFLOW_MLIR_TOOLS	OFF	Build MLIR integration (requires LLVM)
DOMAINFLOW_DSE	OFF	Build design space exploration tools
DOMAINFLOW_VISUALIZATION	OFF	Build visualization tools
DOMAINFLOW_MATPLOT_TOOLS	OFF	Build matplotlib integration
DOMAINFLOW_DATABASE_TOOLS	OFF	Build database tools

Troubleshooting

”Could not find toolchain file: /scripts/buildsystems/vcpkg.cmake”

This happens when CMake cache has stale vcpkg configuration:

# Clean the build directory
rm -rf build/user-ninja-release

# Reconfigure
cmake --preset user-ninja-release

“CMake was unable to find a build program corresponding to Ninja”

Install Ninja:

# Ubuntu/Debian
sudo apt install ninja-build

# Fedora/RHEL
sudo dnf install ninja-build

# macOS
brew install ninja

vcpkg not found (Windows only)

Ensure VCPKG_ROOT environment variable is set:

echo $env:VCPKG_ROOT  # Should print vcpkg path

MLIR not found

If building with DOMAINFLOW_MLIR_TOOLS=ON:

1. Verify LLVM_PROJECT_ROOT is set in CMakeUserPresets.json
2. Ensure LLVM/MLIR is built
3. Pass -DMLIR_DIR explicitly if needed

Preset not found

If you see “preset not found”, ensure you’ve created CMakeUserPresets.json from the appropriate platform template.

Compiler warnings about overflow

The warnings in index_space.hpp about overflow conversions are expected and harmless. They occur when using double::infinity() with integer types for sentinel values.

Platform-Specific Notes

Linux

• Use Ninja-Debug or Ninja-Release presets (or user variants)
• Ensure CMake 3.31+ is installed via system package manager or from source

Windows

• Use VS17-Debug or VS17-Release presets for Visual Studio
• Or use Ninja-Debug/Ninja-Release with Ninja generator
• Install Visual Studio 2022 with C++ development tools

macOS

• Use Ninja-Debug or Ninja-Release presets
• Install CMake via Homebrew: brew install cmake
• Ensure Xcode Command Line Tools are installed