ghadiff 2.1.1

Compare GitHub workflow runs with detailed analysis

ghadiff

A Python CLI tool to compare two GitHub Actions workflow runs with detailed analysis of timing, status changes, and job differences.

Features

• 🔍 Compare any two GitHub workflow runs
• 📊 Detailed job and step-level analysis
• ⏱️ Duration comparisons with percentage changes
• 🎨 Multiple output formats: Text, JSON, Markdown, HTML
• 🚀 Defaults to tenstorrent/tt-metal repository
• 🔒 GitHub API token support with rate limit handling

Installation

From PyPI (once published)

pip install ghadiff

From source

git clone https://github.com/Aswintechie/ghadiff.git
cd ghadiff
pip install -e .

Quick Start

Prerequisites

You'll need a GitHub personal access token for API access:

1. Go to GitHub Settings → Developer settings → Personal access tokens
2. Generate a new token with repo and workflow scopes
3. Set it as an environment variable:

export GITHUB_TOKEN=your_token_here

Basic Usage

Compare two workflow runs (defaults to tenstorrent/tt-metal):

ghadiff 12345678 12345679

The first argument is Run 1 (baseline), the second is Run 2 (comparison).

With Custom Repository

ghadiff 12345678 12345679 --repo owner/repo

Generate Reports

Text format (default):

ghadiff 12345678 12345679

JSON format:

ghadiff 12345678 12345679 --format json

Markdown format:

ghadiff 12345678 12345679 --format markdown -o report.md

HTML format:

ghadiff 12345678 12345679 --format html -o report.html

Output Examples

Text Format

================================================================================
GitHub Workflow Run Comparison
================================================================================

OVERVIEW
--------------------------------------------------------------------------------
Run 1: #1234 (12345678)
  Branch: main
  SHA: abc1234
  Status: completed / success
  Duration: 45.2m

Run 2: #1235 (12345679)
  Branch: main
  SHA: def5678
  Status: completed / success
  Duration: 38.7m

Duration Difference: -6.5m

JOBS COMPARISON
--------------------------------------------------------------------------------
Total jobs compared: 25
  In both runs: 25
  Only in Run 1: 0
  Only in Run 2: 0

build-and-test
   Run 1: ✅ success    - 12.3m
   Run 2: ✅ success    - 10.1m
   Diff:  -2.2m (-17.9%)

...

JSON Format

Full structured data with all workflow, job, and step details for programmatic access.

HTML Format

Beautiful, responsive HTML report with color-coded status indicators and sortable tables.

CLI Interface Example

ghadiff 12345678 12345679 \
  --repo tenstorrent/tt-metal \
  --format html \
  --output report.html

positional arguments: run1 First workflow run ID run2 Second workflow run ID

optional arguments: -h, --help show this help message and exit --repo REPO Repository in format owner/repo (default: tenstorrent/tt-metal) --token TOKEN GitHub personal access token (or use GITHUB_TOKEN env var) --format {text,json,markdown,html} Output format (default: text) --output OUTPUT, -o OUTPUT Output file (default: stdout) --verbose, -v Verbose output (text format only)

## Python API

You can also use the package programmatically:

```python
from workflow_compare import GitHubAPI, WorkflowComparator, Reporter

# Initialize API client
api = GitHubAPI(token="your_token", repo="tenstorrent/tt-metal")

# Fetch workflow runs
run1 = api.get_workflow_run_full(12345678)
run2 = api.get_workflow_run_full(12345679)

# Compare
comparator = WorkflowComparator(run1, run2)
comparison = comparator.get_full_comparison()

# Generate report
reporter = Reporter(comparison)
print(reporter.to_text())

Development

Setup Development Environment

git clone https://github.com/Aswintechie/ghadiff.git
cd ghadiff
pip install -e ".[dev]"

Run Tests

pytest

Code Formatting

black src/

Use Cases

• Performance Regression Detection: Compare workflow runs before and after code changes
• CI/CD Optimization: Identify which jobs got faster or slower
• Debugging Failures: Compare a failing run with a successful baseline
• Release Validation: Ensure new releases don't introduce timing regressions
• Infrastructure Changes: Validate runner or environment changes

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License

Links

• Repository: https://github.com/Aswintechie/ghadiff
• Issues: https://github.com/Aswintechie/ghadiff/issues
• PyPI: https://pypi.org/project/ghadiff/

Acknowledgments

Built for the Tenstorrent tt-metal project to improve CI/CD workflow analysis.