dropshell-build/CLAUDE.md
Your Name 1ac00e83f8
Some checks failed
dropshell-build / build (push) Has been cancelled
'Generic Commit'
2025-06-14 22:07:46 +12:00

129 lines
4.1 KiB
Markdown

# CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
## Project Overview
This repository contains a Docker-based build system for creating statically-linked C++ executables using Alpine Linux and musl libc. It consists of:
1. **dropshell-build-base**: A comprehensive C++ development base image with Drogon web framework and database support
2. **dropshell-build**: A multi-stage Docker build system that uses the base image to compile C++ projects
3. **ipdemo**: A sample application demonstrating the build system
## Key Commands
### Building the Base Image
For local testing (single architecture):
```bash
cd build-base
./build.sh
```
This creates the `gitea.jde.nz/public/dropshell-build-base:latest` Docker image for your current architecture and loads it into Docker.
### Building a Project
```bash
./build.sh
```
This builds the project defined in `build.sh` (default: ipdemo) and outputs the binary to `output/`.
### Testing the Built Binary
```bash
./test.sh
```
Runs the compiled binary from `output/` directory.
### Environment Variables
- `CMAKE_BUILD_TYPE`: Set to "Debug" or "Release" (default: Debug)
- `PROJECT`: The project name to build (default: ipdemo)
## Architecture and Design
### Build System Structure
The build system uses a two-stage approach:
1. **Base Image** (`dropshell-build-base`): Contains all dependencies compiled from source with static linking:
- C++ compiler toolchain with ccache and mold linker
- Libraries: OpenSSL, jsoncpp, PostgreSQL, MariaDB, MySQL, SQLite3, cURL, c-ares
- Drogon web framework with full database support
- All libraries built with `-fPIC` for static linking
2. **Project Build** (`Dockerfile.dropshell-build`): Multi-stage build that:
- Uses build cache mounts for faster rebuilds
- Runs cmake_prebuild.sh for custom pre-build steps
- Generates version information dynamically
- Produces a single static binary in a scratch container
### CMake Configuration
The CMakeLists.txt enforces static linking through:
- `CMAKE_EXE_LINKER_FLAGS` with `-static` flag
- `CMAKE_FIND_LIBRARY_SUFFIXES` set to `.a`
- `BUILD_SHARED_LIBS` forced to OFF
- All libraries installed in standard `/usr/local` paths
Important: Projects should set CMAKE_PREFIX_PATH and may need to explicitly link PostgreSQL libraries:
```cmake
# Set paths for libraries before finding Drogon
set(CMAKE_PREFIX_PATH ${CMAKE_PREFIX_PATH} /usr/local)
# Additional PostgreSQL libraries needed for static linking
set(POSTGRESQL_EXTRA_LIBS
/usr/local/lib/libpgcommon.a
/usr/local/lib/libpgport.a
)
```
### Project Structure for New Applications
New C++ projects should follow this structure:
```
project-name/
├── CMakeLists.txt # Based on ipdemo example
├── cmake_prebuild.sh # Optional pre-build script
└── src/
├── main.cpp # Application entry point
└── version.hpp.in # Version template
```
### Adding New Projects
To build a different project:
1. Create a directory with the project structure above
2. Modify `build.sh` to set `PROJECT="your-project-name"`
3. Run `./build.sh` to build
### Static Linking Details
The system ensures complete static linking by:
- Using Alpine Linux's musl libc
- Building all dependencies with `--enable-static --disable-shared`
- Setting `-fPIC` for position-independent code
- Using mold linker with `-static` flag
- Configuring CMake to prefer `.a` libraries
### Database Support
Applications can use:
- PostgreSQL via libpq
- MySQL via MariaDB connector
- SQLite3
All database libraries are installed in `/usr/local` and are statically linked into the final binary.
### HTTP Client Utilities
The ipdemo includes a blocking HTTP client utility (`src/http_utils.hpp`) that wraps Drogon's asynchronous HttpClient:
```cpp
#include "http_utils.hpp"
// Make a blocking HTTP GET request
auto response = http_get("example.com", "/api/endpoint", true, 10.0); // HTTPS with 10s timeout
if (response.success && response.status_code == 200) {
std::cout << response.body << std::endl;
}
```
This utility handles the threading complexity and provides a simple synchronous interface for HTTP requests.