Code sharing between Scala 2 and Scala 3

[anatoliykmetyuk]

In Dotty, we've encountered a problem that the Dotty's version of ClassfileParser presumably fell behind the Scala 2's one (see scala/scala3#6136). This appears to be a specific case of a more general problem of code sharing between Scala 2 and 3. In this issue, I propose to discuss the ways we can keep the Scala 2 and 3 components DRY and in sync.

I believe the following questions need to be answered:

• What code repeats between Dotty and Scala 2?
• What considerations do we need to take into account when sharing between Dotty and Scala 2? What are the obstacles?
• How do we share the code?

Regarding the second point, I suppose one obstacle would be the coupling it creates between Dotty and Scala 2. The changes to Scala 2 and Dotty may affect each other, and this complicates experimentation.

Another consideration is that the code bases are big, and it may take a lot of work to see what can be shared and to actually share it.

@adriaanm, would love to hear your opinion on the above.

[adriaanm]

I think we should invest in sharing code, and we'll soon have time to start working on this (when 2.13.0 is out). I'm very open to change scalac internals/APIs to make this happen, to reorganize source folders, git submodules,....

Let's start with the backend (genbcode + optimizer), and figure out how to define an interface that works for both frontends with minimal glue code. I suppose it will have to be compiled from source against both frontends for best performance, but if we can find a way to share a binary artifact, that would be amazing.

[SethTisue]

should this stay open? is there anything here that's actionable now from the Scala 2 side?

(I think maybe we should close it and then wait for requests to be initiated from the Scala 3 side?)

[anatoliykmetyuk]

Yes, I believe there are no concrete plans on this. So we can close it.