By popular demand pitest makes it possible to break the build if a certain line coverage or mutation score is not achieved.
Using a metric in this way can however drive the wrong behaviors, resulting in over specified tests and unit tests for code that would be more sensibly integration tested rather than unit tested.
A better metric on which to break the build might be a measure of average test strength - the % of mutants that are executed by a test, but not killed by it.
This could be calculated from the number of killed mutants dived by the number of killed and surviving mutants (excluded those with a status of uncovered).
This ought to provide a metric that falls weak tests are introduced.
By popular demand pitest makes it possible to break the build if a certain line coverage or mutation score is not achieved.
Using a metric in this way can however drive the wrong behaviors, resulting in over specified tests and unit tests for code that would be more sensibly integration tested rather than unit tested.
A better metric on which to break the build might be a measure of average test strength - the % of mutants that are executed by a test, but not killed by it.
This could be calculated from the number of killed mutants dived by the number of killed and surviving mutants (excluded those with a status of uncovered).
This ought to provide a metric that falls weak tests are introduced.