Ensure the Maven and Gradle DependencyInsights recipes treat dependencies coming in through various paths the same

[timtebeek]

Dependencies can come in through various paths; at present we do not show the path of how dependencies come in in the data table rows. This may lead to duplicate rows, which we accept until we add a column to differentiate between rows (not a goal for this PR).

Before

We see a difference between Maven and Gradle DependencyInsight recipes: the Gradle recipe adds identical rows for the varying paths, whereas the Maven recipe does not add those identical rows.

After

Both recipes add rows for each path that might bring in a dependency, even if those rows might be identical.

[github-actions]

Some suggestions could not be made:

• rewrite-gradle/src/main/java/org/openrewrite/gradle/search/DependencyInsight.java
  • lines 22-22

[jevanlingen]

Yep, looking good! I am only wondering about the maven counterpart. If maven's recipe can also have this same problem, I would add it in this PR as well.

[timtebeek]

Adding some context here: the test uses rewrite-core, which indeed has three copies of jackson-core coming in

dependencies {
    implementation 'org.openrewrite:rewrite-core:7.39.1'
}

all three result in the same data table row before the changes made here.

[github-actions]

Some suggestions could not be made:

• rewrite-gradle/src/main/java/org/openrewrite/gradle/marker/GradleDependencyConfiguration.java
  • lines 33-33
  • lines 271-271

[jkschneider]

I’m pretty sure this is a non-goal in fact. Dependencies can come to be in multiple ways in a file. And there is utility in knowing how many times a dependency is included

[timtebeek]

As discussed in Slack: we'll instead change the Maven recipe to match Gradle, as opposed to changing the Gradle recipe to match Maven. We need to update the tests accordingly and the Maven DependencyInsight recipe. In the future we might add a column to distinguish between the data table rows produced.