Reduced mockito usage

[jkronegg]

🤔 What's changed?

I reduced the usage of the Mockito framework. This is a first pass, focused on the simple cases (typically: replacing the mock+when constructs with stub classes).

⚡️ What's your motivation?

Fixes #2707 .

By focusing on the simple cases we have a big pull request of stuff that is simple to review. The other replacements are more complex and prone to more discussions, so I will make another pull request for them.

🏷️ What kind of change is this?

• 🏦 Refactoring/debt/DX (improvement to code design, tooling, documentation etc. without changing behaviour)

♻️ Anything particular you want feedback on?

Performance when running all tests using Intellij (right click on project -> Run -> All Tests ; average duration over 5 runs):

project	test duration (before -> after) [s]	mockito contribution (before -> after)
cucumber-junit	3.3 -> 1.1	2.2 s = 68% -> 0% (removed dependency)
cucumber-core	9.3 -> 6.4	2.9 s = 31% -> 0% (removed dependency)
cucumber-java	2.7 -> 0.8	1.9 s = 72% -> 0% (removed dependency)
cucumber-java8	0.9 -> 0.7	0.2 s = 21% -> 0% (removed dependency)
cucumber-spring	5.3 -> 3.2	2.0 s = 39% -> 0% (removed dependency)
cucumber-testng	1.7 -> 1.7	0.0 s = 0% -> 0% (removed dependency)
cucumber-guice	2.4 -> 0.7	1.7 s = 71% -> 0% (removed dependency)
datatable	0.5 -> 0.5	(unused dependency) 0.0 s = 0% -> 0% (removed dependency)

Thus, when running tests interactively, we see a good improvement by not using mockito (up to 72% of testing time on cucumber-java).

However, over the full build (maven clean install on cucumber-jvm), there is no build time difference: on my machine, the build time is 1 minute 49 seconds on the main branch as well as on branch reduce-mockito-usage-pass1 (average build time over 5 builds).

My guess is that the build time is much longer than the test duration, so the test duration on total maven clean+install is not noticeable.

Edit:
One year after, I did additional tests on my local machine for a maven clean install on cucumber-jvm (5 times average):

• with mockito ( main branch ): average=3 minutes 42 seconds (standard deviation = 8 seconds)
• without mockito (reduce-mockito-usage-pass1 branch): average=3 minutes 26 seconds (standard deviation = 12 seconds)

That is: the build is 7% faster without mockito.

📋 Checklist:

• I agree to respect and uphold the Cucumber Community Code of Conduct
• I've changed the behaviour of the code
  • I have added/updated tests to cover my changes.
• My change requires a change to the documentation.
  • I have updated the documentation accordingly.
• Users should know about my change
  • I have added an entry to the "Unreleased" section of the CHANGELOG, linking to this pull request.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.00%. Comparing base (ba6a53d) to head (f75916b).
Report is 3 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff              @@
##               main    #2767      +/-   ##
============================================
+ Coverage     84.99%   85.00%   +0.01%     
- Complexity     2734     2736       +2     
============================================
  Files           331      331              
  Lines          9559     9559              
  Branches        918      918              
============================================
+ Hits           8125     8126       +1     
  Misses         1109     1109              
+ Partials        325      324       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mpkorstanje]

I'm halfway through reviewing this. Made #2783 to reduce the size a bit.

[jkronegg]

I removed all mockito usages. I'm happy with the performance improvement on interactive tests, but a bit puzzled by having the same build duration with or without mockito when doing a maven clean+install.
Edit:
The build is not 7% faster without mockito, see top post for details.