FormatOps: implement indentOperator.exemptScope

[kitbellew]

current topLevelOnly was likely intended to implement https://github.com/scala-js/scala-js/blob/main/CODINGSTYLE.md#long-expressions-with-binary-operators but the result is not as described.

the new enclosedAlone option attempts to do that instead, but needs confirmation from scalajs developers.

[kitbellew]

@ekrich @sjrd please comment

[sjrd]

There is one thing below that is off compared to our style guide. Otherwise it looks perfect. :)

[sjrd]

Theses would also be

Suggested change

	foo(
	a &&
	b,
	a &&
	b
	)
	foo(
	a &&
	b,
	a &&
	b
	)

Even when there are multiple parameters, the binary operator chain is alone in their actual parameter.

[kitbellew]

what's an actual parameter?

[kitbellew]

and if you don't need them to be enclosed (which one parameter gives you but multiple do not), why is being alone in the case body (which is what started this conversation) not included?

[sjrd]

An actual parameter is an argument at call site. As opposed to a formal parameter, which is the parameter at definition site.

and if you don't need them to be enclosed (which one parameter gives you but multiple do not), why is being alone in the case body (which is what started this conversation) not included?

Hum actually being alone in a case body should be included as well, yes.

The "alone" condition is not really about being enclosed. It's about avoiding things like

val foo = {
  doSomethingFirst()
  thenReturnThis ||
  that
}

because there it's extremely confusing.

[kitbellew]

ok, now I'm confused. @ekrich, didn't you say that the recent change to infix as entire case body was not welcomed?

@sjrd any other cases? so it's not about having delimiters but being alone? then what about

if (...)
  a &&
  b

indent or not? it's alone in then body. or method body? or yield body?

or

// infix retrieves a lambda, and it's alone
(foo get
  bar)(arg)

[sjrd]

@sjrd any other cases? so it's not about having delimiters but being alone? then what about

if (...)
  a &&
  b

indent or not? it's alone in then body. or method body? or yield body?

Our coding style would mandate {} around the then branch (because it has more than one line; yes I know scalafmt won't do this for me ;-) ), so this is a case that cannot happen for us. If I go beyond our own style, and we accept this as valid, I think it should be considered alone in the then body, yes. I think this is only way that it can decently generalize to braceless syntax anyway.

or

// infix retrieves a lambda, and it's alone
(foo get
  bar)(arg)

Good question. I don't think I would get in that situation, given my coding habits. I don't have a good answer for this one. I would say do whatever is simpler to implement. 🤷‍♂️

[kitbellew]

Our coding style would mandate {} around the then branch (because it has more than one line; yes I know scalafmt won't do this for me ;-) )...

i implemented this rewrite recently: https://scalameta.org/scalafmt/docs/configuration.html#inserting-braces

[tgodzik]

Looks good from my perspective, though I have no stake at any particular style here.