We found the need to have a particular kind of in-memory cache to cache the result of an operation that involved an expensive DB call plus some computation. We wanted the cache to:
- be kept fresh, either by refreshing on schedule or on an event
- be thread-safe
- not to block whilst refreshing (an IO-bound operation in our case)
This is purely in-process and probably has a narrow range of use-cases. It is probably also re-inventing the wheel - we'll surely stumble across another, better implementation before long!
To use you'll want to sub-class ActiveCache and implement refresh_cache which computes and returns the value to cache, and possibly trigger which blocks and returns only when you wish the cache to refresh.
TimeoutCache is an implementation which refreshes the cache on a fixed interval. By sub-classing and implementing refresh_cache you will have a cache that refreshes on a periodic basis.
Read the code comments for more info.
With pip:
> pip install activecache
From a GIT checkout:
> python setup.py install
If you have a GIT checkout of this repository you can run the tests as follows:
- create and activate a virtualenv for the project
- pip install -r test_requirements
- ./runtests.sh
This will run all the unit tests, create a coverage report (see htmlcov/index.html) and also PEP8 check the code (see pep8.txt).
If you like to use Jenkins you may like to make use of our build script build.sh. If executed by Jenkins as a build step you can then ingest the output as follows:
- Cobertura plugin can read coverage.xml
- xUnit plugin can read JUnit formatted nosetests.xml
- violations plugin: add pep8.txt to the pep8 field
Copyright |copy| 2012 ReThought Ltd, all rights reserved