Implement the rule to check if dispatchers are injectable

[chao2zhang]

This resolves #4175.

[codecov]

Codecov Report

Merging #4222 (11a5a8c) into main (c848341) will increase coverage by 0.12%.
The diff coverage is 82.60%.

@@             Coverage Diff              @@
##               main    #4222      +/-   ##
============================================
+ Coverage     84.19%   84.32%   +0.12%     
- Complexity     3233     3249      +16     
============================================
  Files           468      469       +1     
  Lines         10185    10186       +1     
  Branches       1786     1783       -3     
============================================
+ Hits           8575     8589      +14     
+ Misses          671      658      -13     
  Partials        939      939              

Impacted Files	Coverage Δ
...rbosch/detekt/rules/coroutines/InjectDispatcher.kt	81.81% <81.81%> (ø)
...osch/detekt/rules/coroutines/CoroutinesProvider.kt	100.00% <100.00%> (ø)
...les/style/ExplicitCollectionElementAccessMethod.kt	68.96% <0.00%> (-14.37%)	⬇️
...rbosch/detekt/rules/style/ObjectLiteralToLambda.kt	83.78% <0.00%> (-3.40%)	⬇️
...turbosch/detekt/rules/style/ForbiddenMethodCall.kt	87.80% <0.00%> (-3.31%)	⬇️
.../io/gitlab/arturbosch/detekt/rules/style/NoTabs.kt	92.30% <0.00%> (-2.43%)	⬇️
...detekt/rules/bugs/DoubleMutabilityForCollection.kt	85.18% <0.00%> (-0.53%)	⬇️
.../kotlin/io/gitlab/arturbosch/detekt/cli/CliArgs.kt	100.00% <0.00%> (ø)
...ain/kotlin/io/github/detekt/metrics/LinesOfCode.kt	95.45% <0.00%> (ø)
...rturbosch/detekt/rules/style/UnusedPrivateClass.kt	76.59% <0.00%> (ø)
... and 14 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c848341...11a5a8c. Read the comment docs.

[cortinico]

Great stuff :)

[BraisGabin]

I don't understand the use case to allow to configure this? Shouldn't we just check all the usages of Dispatchers.*? If this is because you can change the Main and the Immediate maybe a boolean flag (disable by default) could be a better fit for discoberavility.

[chao2zhang]

Dispatchers.Main is generally okay because we have Dispatchers.setMain()/resetMain() available in test (https://kotlin.github.io/kotlinx.coroutines/kotlinx-coroutines-test/)

Additionally, if users define their own Dispatchers, this list is necessary for specifying their customized dispatcherNames.

[BraisGabin]

Dispatchers.Main is generally okay because we have Dispatchers.setMain()/resetMain()

It's possible to change them, but I think that adding Main is a safer default. If someone is OK using those function it can remove it. This kind of test functions aren't very good. If you forget to use resetMain you can change the behaviour of other tests.

For example JodaTime had DateTimeUtils.setCurrentMillisFixed but it was removed in java.time. Now you are forced to inject a Clock if you want to test it. I think that we should add Main but default for the same reason.

[chao2zhang]

Jetbrain announces its plan to improve kotlin coroutin test https://youtrack.jetbrains.com/issue/KT-46783.

On the other hand, [Dispatchers.Main] is used in common androidx libraries:

• Paging3: https://cs.android.com/androidx/platform/frameworks/support/+/androidx-main:paging/paging-common/src/main/kotlin/androidx/paging/PagingDataDiffer.kt?q=Dispatchers.Main&ss=androidx
• LiveData: https://cs.android.com/androidx/platform/frameworks/support/+/androidx-main:lifecycle/lifecycle-livedata-ktx/src/main/java/androidx/lifecycle/CoroutineLiveData.kt?q=Dispatchers.Main&ss=androidx&start=61
• Lifecycle: https://cs.android.com/androidx/platform/frameworks/support/+/androidx-main:lifecycle/lifecycle-runtime-ktx/src/main/java/androidx/lifecycle/RepeatOnLifecycle.kt?q=Dispatchers.Main&ss=androidx&start=91

Since most UI applications do have a single Main thread, we are more likely to get complaints if we add Main as the default.

[BraisGabin]

I don't completely agree but we could move this to an issue. There's no need to block the merging for this.