A potential regression in v0.7.0: tlPrePrBotHook produces an empty diff but scalafix brings some changes

[danicheg]

I'm not sure if this is a bug or some changes in behaviour brought in v0.7.0. So I apologise in advance!
Details are here http4s/sbt-http4s-org#260 (comment).

[armanbilge]

Hm, are you sure its a regression? Did it used to work differently?

tlPrePrBotHook currently doesn't run Scalafix. I think I decided against this because running Scalafix requires compiling the entire repository and I was afraid to ask Steward to be doing so much work.

sbt-typelevel/core/src/main/scala/org/typelevel/sbt/TypelevelPlugin.scala

Lines 89 to 91 in 9371b95

	val botHook = List("githubWorkflowGenerate") ++
	List("+headerCreateAll").filter(_ => header) ++
	List("+scalafmtAll", "scalafmtSbt").filter(_ => scalafmt)

[danicheg]

Hm, are you sure its a regression? Did it used to work differently?

In the beginning, it was kind of 50% certainty, for now, it's like 10% 😆
I was always thinking that prePR runs everything from formatting code/headers to Scalafix. If not, how do we set up running Scalafix in CI with sbt-typelevel then?

[armanbilge]

I was always thinking that prePR runs everything from formatting code/headers to Scalafix.

Yes, that's correct. This one is for users to run on their local machine.

But tlPrePrBotHook is a "lighter weight" version because I was afraid to burden the Steward with too much work to compile every repository for every sbt-typelevel update. But I just remembered that Steward already runs some scalafix rules sometimes. If those are working in the same way, then maybe it not as much of a problem as I feared.