`javancss-maven-plugin` fails on java 8+ codebase

[fbiville]

maven-project-info-reports-plugin 3.0 is not on Central, 3.0.0 is.
It's blocking mvn site from completing.

[fbiville]

Here is what I had to change in Assertj-Neo4j to get the site to generate:

1. remove all Java 8 (luckily, was just 3 lines of code to change)
2. override the following plugins:

+    <pluginManagement>
+      <plugins>
+        <plugin>
+          <artifactId>maven-site-plugin</artifactId>
+          <version>3.7.1</version>
+        </plugin>
+      </plugins>
+    </pluginManagement>
   </build>

+  <reporting>
+    <plugins>
+      <plugin>
+        <groupId>org.apache.maven.plugins</groupId>
+        <artifactId>maven-project-info-reports-plugin</artifactId>
+        <version>3.0.0</version>
+      </plugin>
+    </plugins>
+  </reporting>

[joel-costigliola]

So all we need to do is to bump the version of these plugins ?

[fbiville]

It seems the above changes work well for pre-Java 8 code.
If you got lambdas, method references etc... there will still be parsing errors.

[joel-costigliola]

@fbiville we have just released assertj-parent-pom 2.2.4, can you give it a try to see if it solved your issue ?

[fbiville]

@scordio fixed it joel-costigliola/assertj-neo4j@2667c3f :)

[fbiville]

Oopsie, talked too fast:

cd $(mktemp -d) && \
    git clone https://github.com/joel-costigliola/assertj-neo4j.git && \
    mvn site -f assertj-neo4j

[fbiville]

Now that Assertj Neo4j contains Java 8+ code again, the site cannot be generated.

[joel-costigliola]

java 8 ? living on the edge @fbiville ? 😆

[scordio]

The missing support for Java 8 seems to be a JavaNCSS 33.54 issue used by javancss-maven-plugin. There is a pull request for it but no feedback for more than a year: codehaus/javancss#3

The fork from Nokia should have already the fix but I didn't find the updated artifact deployed anywhere.

As the only plugin I see for reporting is maven-project-info-reports-plugin, I guess it's the one responsible invoking the javancss-maven-plugin. Would it be an option to exclude the JavaNCSS report from its execution? (Not sure yet how to do it, I'll investigate a bit more later on)