I will implement adoption to this feature for Gradle Plugin in the separated PR

OK. So in your next PR you will just make use of the new generated java routes for the gradle plugin or did you plan to also enable it for the sbt java plugin?

If you want you can also already submit the next PR (including the commit of this PR). Just so I can take a look.
I can still review them separate and merge them separate.

So in your next PR you will just make use of the new generated java routes for the gradle plugin or did you plan to also enable it for the sbt java plugin?

Only for Gradle plugin. Yes, I thought about enable a java routes by default for PlayJava sbt plugin, but decided don't do that before we'll get a feedback from someone who enable it manually. If you think we are ready do that right now, I will be happy do that.

A lot of changes in this file can be combine and optimise for both languages. But I decided keep it to the future yet and avoid touching an exists functions for Scala and wrote a "duplicate" for Java.
@mkurz if you want, I can try to do this unification in this PR.

javaJava?

Yep. It looks not good, but It's just ajava prefix for all existed templates.

Hmm.. yes it's a bit confusing to read.

Just thinking out loud, maybe we should also rename existing template files:

jJavascriptReverseRouter.scala.twirljJavaWrappers.scala.twirljReverseRouter.scala.twirljRoutesPrefix.scala.twirl

sJavascriptReverseRouter.scala.twirlsJavaWrappers.scala.twirlsReverseRouter.scala.twirlsRoutesPrefix.scala.twirl

and

sForwardsRouter.scala.twirljForwardsRouter.scala.twirl

But not sure, I think prefixing existing templates withs will break compatibility. So maybe just prefix the new templates just withj instead of writing outjava. We do use this approach with some methods and variables in the code, maybe also here? What do you think?

I think the names of templates are internal part and we are free to choice any pattern for them.
I like an idea with one letters/j prefix. 👍

Without this filternot the simple example

GET     /  controllers.HomeController.index(r: Request)

publicResultindex(Http.Requestr)  { ... }

fails. Becauseps.mkString("", ",\n ", ",") above will add the trailing, even if the list is empty, so we end up withcall(, ...) which of course does not compile (This is more or less the equivalent of the.filterNot(_ == "()") in the scala method).

OMG, it's so stupid checked a lot of cases with any type of params and forgot about case with only oneRequest param 🤦‍♂️

Test for this case was added in this commit99aad10

I think we can unify[java]JavaWrapper.scala.twirl (specially because even in play scala it will be generated as java file).
I think we can just stupidly wrap the scala reverse routes in aReverseRoutes class/object and also pass the prefix lamba as lambda and not lazy string.
With that two changes we should be able to just use one file.
Feels a bit cleaner IMHO as the differences between[java]JavaWrapper.scala.twirl is not really big.

I'm not sure that I understood your idea for 100%, but I unifiedroutes.java for both compilers. Can you check that (separated commit)?

I think we can just stupidly wrap the scala reverse routes in aReverseRoutes class/object

Looks like If someone uses a reverse routes classes directly then these changes will break a source compatibility in new version. Is it ok to you?