add test case to illustrate issue with unsolicited indentation.

[timo-a]

What's changed?

This PR adds a test to illustrate a bug.

What's your motivation?

I want to enforce a custom order of imports on new PRs through a custom style.
As I report in "observation 2" of #4165 custom styles also change indentation. I don't think they should do that.

Anything in particular you'd like reviewers to focus on?

Anyone you would like to review specifically?

Have you considered any alternatives or workarounds?

Any additional context

I've marked this PR ready for review in he hope of triggering a build, but it seems a manual approval is necessary.
I won't be available to work on a fix.

Checklist

• I've added unit tests to cover both positive and negative cases
• I've read and applied the recipe conventions and best practices
• I've used the IntelliJ IDEA auto-formatter on affected files

[Laurens-W]

Hey @timo-a, I think in your case you also want to influence the tabs and indents like we do here in our IntelliJ style.
Specifically the 4th parameter of the TabsAndIndentsStyle here influences the continuation indent

[timo-a]

Thanks for the pointer! I expected that not defining a TabsAndIndentsStyle in the first place would leave every indent as it is but apparently not. When I set the 4th parameter continuationIndent to

• 4 the test works, but then every continuation is indented by a fixed value which is not what I want
• null the apparent default value of 8 is used
• 0 there is no indentation at all
• -1 the token is lifted to the line before
• -2 or less the original indentation is kept.

So, with some configuration, namely setting continuationIndent to -2 it can be done already. Unfortunately this behavior is not documented at all and I worry that it will be "fixed" in the future, so I modified my test to confirm it. Please merge it in and do consider adding javadoc to at least continuationIndent.

[Laurens-W]

@timtebeek is currently unavailable until Tuesday, Let's wait for his response on whether this is expected behavior and if it is we can merge the PR as regression test 👍

[knutwannheden]

Adding something like this to the TabsAndIndentsVisitor should fix this specific case, although the same should then probably also be done for JContainer and JRightPadded and by extension then possibly also for some other formatter visitors:

    @Override
    public <T> JLeftPadded<T> visitLeftPadded(@Nullable JLeftPadded<T> left, JLeftPadded.Location loc, P p) {
        if (getCursor().getNearestMessage("stop") != null) {
            return left;
        }
        return super.visitLeftPadded(left, loc, p);
    }

Background: While the stopping mechanism works, it short-circuits any calls to visit() and thus prevents the visitor from descending down the tree. However, the spaces (which is what the formatter changes in the end) can also be directory contained by JRightPadded, JLeftPadded, and JContainer and there is nothing stopping them from being visited.