Lightweight service framework on top of gevent, implementing the "service model" -- services all the way down.
- Service model -- Break your app down into services and sub-services. Modules, if you will, that can start, stop, and reload. Every service manages its own pool of greenlets.
- Configuration -- Built-in, reloadable configuration based on Python files. Access configuration settings relative to services.
- Runner -- Command-line tool to manage your service that can daemonize, chroot, drop privs, and set up or override configuration.
A talk was given at PyCon 2012 called "Throwing Together Distributed Services with gevent" that used Ginkgo to build a number of simple services combined to make a more complex distributed service.
Pretty active discussion on this early microframework. Join it or just read what's being planned:
Feel free to poke around the issues in the main repository and see if you can tackle any. From there you should:
- Fork if you haven't
- Create a branch for the feature / issue
- Write code+tests
- Pass tests (using nose)
- Squash branch commits using merge and reset
- Send pull request
We highly recommend using branches for all features / issues and then squashing it into a single commit in your master before issuing a pull request. It's actually quite easy using merge and reset. This helps keep features and issues consolidated, but also makes pull requests easier to read, which increases the speed and likelihood of being accepted.
We're aiming for at least 90% test coverage. If you have the
coverage Python package installed, you can run
python setup.py coverage to get a coverage report of modules within gservice.