Selection pattern for source classes and test classes
nrainer opened this issue · 3 comments
I am wondering if there is a reason why PmpContext#getClassesInternal creates patterns for packages. Is this optimization needed?
Patterns using the full path to the Java files might be more precise for projects in which the package does not start with the name of the project (so that packages with the same name can exist in multiple projects).
If there is no reason against it, I think this is sensible and can reduce unnecessary analysis efforts.
For example, in the project Achilles, integration tests of different modules reside in the same package so that they would be executed in each integration test module, causing a large overhead. (https://github.com/doanduyhai/Achilles)
Because it is what PIT does.