[ARCHETYPE-584] don't normalize whitespace in tests so we can test for whitespace

[elharo]

@newur @Tibor17

Passes in Java 8. I want to see if this fails in Java 11.

[newur]

Looks like it fails on JDK 11+

https://ci-builds.apache.org/job/Maven/job/maven-box/job/maven-archetype/job/linebreak/1/testReport/

[elharo]

Yep:

Expected: Expected text value '
  ' but was '
    
  ' - comparing <project ...>
  </project> at /project[1]/text()[1] to <project ...>
    
  </project> at /project[1]/text()[1]:
<project>
  <packaging>pom</packaging>
<modules>  <module>myArtifactId1</module>
  </modules>
</project>
     but: result was: 
<project>
    
  <packaging>pom</packaging>
  
  <modules>
      
    <module>myArtifactId1</module>
      
  </modules>
</project>

[elharo]

It might make sense to create a new Java 11 test rather than changing this one.

[newur]

@elharo

I don't see what you try to achieve or what your reasoning is here. However, the call to .normalizeWhitespace() is rather new and I assume it was added to 'get the test green'. So undoing the normalize is bringing the test back to the status quo - which was a stronger test anyway!

You can see it here: https://github.com/apache/maven-archetype/commit/bf7961805ea56cdad7e138f47098aacccb314db8#diff-db4ddee9a40aadf3fab866c350ca950397f69b630de05a126f0265e2136cc070

[elharo]

I'm trying to understand what's going on here, and how it fails. This is easier to do one change at a time. This test isn't very well factored to start with, and might need some cleanup before the issue can be fixed.

[newur]

I'm trying to understand what's going on here, and how it fails.

Right, that is always a good approach to solve a bug.

I would hope that the existing PR is a reasonable starting point on how to fix the bug and my comment in JIRA points to the commit that most likely causes the bug.

Anyway, I wish you good luck with the approach you apply to the problem. :)

[elharo]

I think you've identified the likely cause of the bug. I do not, however, agree with the proposed fix. There's nothing about this bug that suggests we should be adding schema processing. I'll be quite surprised if that turns out to be necessary.