Store less data #22
Description
The tracker quickly overwhelms the database with a lot of data. The data might be useful, but we aren't in a place where we can process it easily. Instead of having to manually delete this old data, the tracker should store less data.
Here's what I'm imagining:
For each unexpected result store the following:
- Test path
- Unexpected type (fail expected pass, pass expected fail, subtests differ)
- Count
- Time
When we find a duplicate (test paths and unexpected type match), we simply increment the count and update the time. We will not have detailed information about subtest failures, but we cannot process this information yet anyway. Later, when we can do something with it, we can re-add this functionality.