Creating partial tests
About partial tests
Normally, dcover create
will create complete tests for a method. A complete test consists of Arrange, Act, and Assert sections and passes when executed. However, sometimes Diffblue Cover is not (yet) able to write complete tests and will give the reason for not creating complete tests in the output summary. In these cases, you can opt to retrain the partial tests created by Cover using the --keep-partial-tests
option. Partial tests may be incomplete in various aspects:
The test does not have assertions.
The test does not always pass when executed.
The Arrange or Act sections produce an error when executed.
The test may execute code that is potentially harmful to your system, leaks resources, or times out.
Creating partial tests is useful to developers for two reasons:
Firstly, it will save developers time. Instead of writing tests from scratch, they'll be able to use the partial tests as a starting point.
Secondly, it may help developers understand why Diffblue Cover was not able to complete the test automatically, especially in cases where the reason for not producing a test is lack of testability.
Note that manually modified tests in test class files that are controlled by Diffblue (e.g. those suffixed with DiffblueTest
) may be overwritten when running Cover the next time.
For example, for method increment
in the following class:
Using dcover create --keep-partial-tests
, Cover will create a partial test:
The problem is clearly that Diffblue Cover has no opportunity to assert on the side effect of the increment
method on field x
, which is marked private
. The developer could add a getter to MyClass
:
Running dcover create
again (with or without --keep-partial-tests
) would now create a complete test.
Last updated