The next alpha of Coverage.py 5.0 is ready: 5.0a2. The big change is that instead of using a JSON-like file for storing the collected data, we now use a SQLite database. This is in preparation for new features down the road.
In theory, everything works as it did before. I need you to tell me whether that’s true or not. Please test this alpha release. Let me know what you find.
If you try it, and it works, let me know! Email is good.
If you see a problem, do this:
- First create a reproducible scenario, something that I can recreate.
- Try running that scenario with the environment variable COVERAGE_STORAGE=json defined, which will use the old JSON storage format. It will be very helpful to know if the results change.
- Write up the issue on GitHub. Please provide as many details as you can.
The biggest change in behavior is that the data file is now created earlier than before. If you are running tests simultaneously, this might mean that you need parallel=true where you didn’t before. Keep an eye out for that.
Some other notes about these changes:
- For now, the old JSON storage code is still in place. It can be enabled with a COVERAGE_STORAGE=json environment variable.
- But I would rather not keep that code around forever. One of the things I’m trying to find out with this alpha is whether there’s any reason I will need to keep it around.
- The database schema is still in flux, and will need to change to support future features. I’m not sure whether to make the schema part of the public interface to coverage.py or not. I want people to be able to experiment with the collected data, but I also want to be able to change it in the future if I need to.
Please test the code, and let me know what you find.