Apply inspections

[mdedetrich]

This is a PR that applies a large set of inspections/syntax/formatting changes. In addition since next release of pekko-connectors will be a breaking change, this opens up additional improvements we can make which are noted below

• Every case class has been made final unless it causes the Scala 2 specific The outer reference in this type test cannot be checked at run time. error (this has been solved in Scala 3 but its not worth splitting the code base over this). This is done because idiomatically in Scala, case class's should always be final because if you allow users to extend a case class they can break properties of case class that you rely on (i.e. you can break structural equality by overriding equals). This shouldn't cause any real world problems, and if it does (i.e. someone is extending by subclassing one of our case classes I would argue this is a good thing as we should be aware of why someone is doing this.
• For custom extensions i.e. extending ExtensionId explicit types have been added to the overridden lookup methods which are more specific/correct than leaving out the explicit type ascription. This is needed because if you don't apply a type to the lookup method then it just defaults to ExtensionId[_ <: Extension] which less specific/useful than lets say GoogleExt.type (which extends ExtensionId[GoogleExt] and ExtensionIdProvider). MiMa filters have been added to help this.
• private/internal class has been made final and/or private where it works to do so

[He-Pin]

I would suggest commit one by one.

[He-Pin]

Another thing is case class can be extended in Java, this is not binary compatible

[mdedetrich]

Another thing is case class can be extended in Java

It doesn't matter if it's Java or Scala, case classes shouldn't ever be extended because the usage of case class implies certain properties that can be broken if a user overrides them.

If a case class is intended to be overridden, you should using a normal class instead.

this is not binary compatible

The entire Pekko 1.1.x series is not binary compatible with Pekko 1.0.x, that's why this is being done now

[He-Pin]

I think we should add annotations. @INTERNAL_API and/Or @doNotInherent in 1.1 cross the projects, and mark it final in 1.2 or 2.0 what ever.

In pekko, there are some classes which is not annotated with these annotations but just comments internal api.sorry, typing from phone

[mdedetrich]

I think we should add annotations. @INTERNAL_API and/Or @doNotInherent in 1.1 cross the projects, and mark it final in 1.2 or 2.0 what ever.

In pekko, there are some classes which is not annotated with these annotations but just comments internal api.sorry, typing from phone

Alpakka has never had any prpper binary compatibility guarantees, it's simply not practical with a monorepo with 20+ integrations and we inherited that.

Due to the fact that we have done major underlying library updates causing breaking changes to many connectors already, the entire release will be classed as breaking.

The plan is that once that release is made, we will start splitting off the connectors into their own repos so we don't get the issues stemming from a monorepo

[He-Pin]

Okay, but we should add those annotations in pekko core project

[samueleresca]

I had a quick skim through and left few minor comments. Is this change a good candidate for the .git-blame-ignore-revs?

[mdedetrich]

@samueleresca Thanks, I have just force pushed with all of the changes fixed. I guess that technically speaking some of the changes can be added to .git-blame-ignore-revs but I would have to manually seperate the pure syntax changes with the non syntax ones and that would take a bit.

I can do this once all of the other issues with the PR have been resolved.

[mdedetrich]

Okay, but we should add those annotations in pekko core project

We can do this once we start splitting off the connectors into their own repo's as it will make sense then. The idea is that this release of pekko-connectors will the last of the "massive breaking changes"

[mdedetrich]

FYI don't merge this PR, I am in the process of splitting out the pure formatting changes into a separate commit so I can add it to .git-blame-ignore-revs and also cherry pick it into 1.0.x, stay tuned!