I have another question: I compile a Scala project with

-Wunused:imports

and there are a few cases where I need to put unused imports into the code to make it compile with pants. Is this a bug or do I have to add some extra metadata to help the dependency detection? It happens with classes that are in the same package and thus normally don't need explicit imports.

These are failures of the Pants dependency inference. If you are able to isolate the source code context in which the added imports are used , then we can fix the Pants Scala parser to account for them.

https://gitlab.com/jberkel/digero/-/blob/73381acfc27f31add7e790099fa74ed3197d75c9/core/src/main/scala/digero/ProcessXMLDump.scala The imports in question are on line 3. The classes are used in the pattern matching on lines 22 + 32.

Can you open an issue?

Sure

Instead of adding unused imports in the code to guide dep inference, you can add to

dependencies=[]

in the BUILD file target

Sorry, I'm struggling a bit. In which BUILD file would this have to go? In the toplevel one, or do I have to create a new one at the level where the dependency is located?

Some BUILD file contains a scala_sources() target A whose deps are being incorrectly inferred. So you'd do something like:

scala_sources(
  overrides={
    "my_source.scala": { "dependencies": ["path/to/my/missing/dep.scala"]}
  }
)

I missed the

overrides

, works now! 👍