Conflict between new line before / after implicit and verticalMultilineAtDefinitionSite?

[alexarchambault]

• Version: 2.0.0-RC4
• Integration: CLI
• Configuration:

maxColumn = 80
continuationIndent.callSite = 2
continuationIndent.defnSite = 2
align.openParenCallSite = false
align.openParenDefnSite = false
danglingParentheses = true

verticalMultiline.newlineBeforeImplicitKW = false
verticalMultiline.newlineAfterImplicitKW = true
newlines.afterImplicitKWInVerticalMultiline = true

verticalMultilineAtDefinitionSite = true

Steps

Given code like this:

case class Foo(
  a: Int,
  b: Boolean,
  c: String,
  d: String,
  e: String,
  f: String,
  g: String
) {
  def run[F[_]](
    fetch: ResolutionProcess.Fetch[F],
    maxIterations: Int = ResolutionProcess.defaultMaxIterations
  )(implicit
    F: Monad[F]
  ): F[Resolution] =
    ???
}

Problem

Scalafmt formats code like this:

case class Foo(
  a: Int,
  b: Boolean,
  c: String,
  d: String,
  e: String,
  f: String,
  g: String) {
  def run[F[_]](
    fetch: ResolutionProcess.Fetch[F],
    maxIterations: Int = ResolutionProcess.defaultMaxIterations
  )(implicit
    F: Monad[F]
  ): F[Resolution] =
    ???
}

Expectation

I would like

  f: String,
  g: String
) {

to be left as-is, rather than changed to

  f: String,
  g: String) {

Workaround

Setting

verticalMultilineAtDefinitionSite = false

fixes that, but in turn, changes

  )(implicit

to

  )(
    implicit

(which I would like to avoid too).

[olafurpg]

Thanks for reporting! The two features were contributed separately so it could be that they don't play nicely together.

[kitbellew]

@alexarchambault you have reported two issues:

• dangling parens in case of vertical multiline
• split before implicit outside of vertical multiline

the first issue should be solved by setting verticalMultiline.excludeDanglingParens = [] (which, for some reason, include class and trait by default). the vertical multiline logic doesn't use the danglingParentheses setting (which also only applies to methods).

the second issue remains, and i will send in a patch to address it.

[alexarchambault]

Thanks a lot for having fixed that, @kitbellew!