I maintain a bunch of Open Source projects. Most of them in Python. Most of them on GitHub.
I find it hard to keep track of when a project has accumulated enough important changes and it's time to make a release.
This is a script to help:
- it runs as a Jenkins job every hour
- re-uses my Jenkins workspaces to look at git clones of the projects (they're in /var/lib/jenkins/jobs/*/workspace)
- finds the latest tag in each Git repo and counts commit since them
- collects some other data from the public GitHub API (such as the number of open issues) and other sources (such as Python version support classifiers in setup.py).
- generates an HTML page with all this information
You can see the result at https://projects.gedmin.as/
- set up a Jenkins job to build this hourly (make && bin/summary --html -o index.html)
- create /var/www/projects.gedmin.as/
- symlink /var/lib/jenkins/jobs/project-summary/assets and index.html into /var/www/projects.gedmin.as/
- set up Apache to serve /var/www/projects.gedmin.as at http://projects.gedmin.as/
Can anyone else use this?
Probably! Don't hesitate to file bugs (or pull requests) asking for more configurability.
Currently all the configuration is hardcoded near the top of
repos.txt. It should be moved to a config file.
Note on HTTP request caching
HTTP requests are cached for 15 minutes by default, in an SQLite database
This is because I run the script rather often while I'm developing it, and without caching I'd run into GitHub's public API rate limits (60 requests per hour) very quickly. The default cache duration lets me have up to 15 projects, which is exactly the number that I have. ;)
You can change the cache duration by specifying, e.g.
(valid units are seconds, minutes and hours and can be abbreviated to
sec/min/hour or s/m/h).