GeezerVison

@tbroyer Thank you very much for your contribution!

The only thing I'm questioning is, given how many parameters GoogleJavaFormatStep::create has now, if we'd benefit from replacing its new variation with a builder. @tbroyer @nedtwigg What do you think?

If we decide the answer is "no", then I wonder if it would be best for the boolean parameter to come before the Provisioner, as provisioners are a Spotless concept, and moving it to the right of the other parameters might be easiest to understand.

Other than that, this PR LGTM, so well done and thanks again!

I agree with @jbduncan that this would be easier to read if we had a builder. But because the list of parameters is so long, I think it's hopeless to worry about the parameter positions. I'm going to merge and release as-is, but a future refactor towards a builder would be welcome. My guess is that this will be the last feature anyone is able to add before they're forced to confront the builder mess.

Released in plugin-gradle 5.15.0 and plugin-maven 2.13.0

Fwiw, tests aren't actually run with JDK 8 on CI ; causing a risk of regressions that won't be detected. Not sure how much you care about it.

I think we test JDK 8, but windows-only. Agreed that it's a bit hidden.

We get enough flaky dep-resolution problems that I don't want to trigger manual retries for every cell in the win/unix vs 8/11/latest matrix.

Strange, this test should have failed for months if that was the case: ecca2b8