* fix: update Go CI workflow and fix formatting issues
- Update Go version from 1.21.x to 1.24.x to match go.mod requirements
- Fix Go module cache path to use linux/go.sum instead of **/go.sum
- Run gofmt on all Go files to fix formatting issues
- Fix benchmark files formatting
- Fix linux/pkg/api/server.go formatting
This resolves the GitHub Actions CI failures related to:
- Missing go.sum file (wrong cache path)
- Go version mismatch
- Code formatting violations
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* fix: add platform-specific syscall.Select wrappers for Linux/Darwin compatibility
- Create select_linux.go: handles syscall.Select returning (n int, err error)
- Create select_darwin.go: handles syscall.Select returning (err error)
- Update select.go to use platform-agnostic selectCall function
- Resolves typecheck errors while maintaining compatibility on both platforms
Tested on both macOS and Linux targets successfully.
---------
Co-authored-by: Claude <noreply@anthropic.com>
* Add comprehensive VibeTunnel protocol benchmark tool
Features:
- Complete HTTP API client implementation for VibeTunnel protocol
- Session management benchmarks (create/get/list/delete operations)
- SSE streaming performance testing with latency measurements
- Concurrent user load testing with realistic simulation
- Support for custom hostname/port configuration
- Detailed performance statistics and reporting
Commands:
- session: Test session lifecycle performance
- stream: Benchmark SSE streaming latency/throughput
- load: Concurrent user load testing
Tested against both Go (port 4031) and Rust (port 4044) servers.
Tool successfully creates sessions and measures performance metrics.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Add configurable benchmark comparison tool
Features:
- Compare command with 10-100 configurable runs
- Cross-server API compatibility (Go/Rust fields)
- Session management and streaming benchmarks
- Performance analysis and winner detection
- Comprehensive statistics and throughput metrics
Results show Go server ~27-50% faster than Rust for session operations.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Remove API compatibility shims for clean unified format
- Unified SessionConfig to use Rust API format (command, workingDir)
- Updated all benchmark commands to use clean API
- Go and Rust servers now use identical API format
- 100-run test shows near-identical performance (1% difference)
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Enable high-load testing up to 1000 runs
- Increased run limit from 100 to 1000 for stress testing
- Reveals resource exhaustion under extreme load
- Go server: 203/1000 success in first round, then failures
- Rust server: immediate failure under high load
- Both servers need connection pooling/rate limiting
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
