James Judd

I have a branch that has this working. I'll try to get it cleaned up, passing tests, and in code review next week.

Once we repro and fix adding a test for this would be awesome. Our proto test coverage at the moment is not great.

That would be awesome! I'm happy to help.

As far as I know, strict mode for rules_scala does not yet catch extra dependencies. @ittaiz discussed writing a compiler plugin to deal with this part. I ran across this...

Well, I feel silly. I didn't read closely enough. Looks like that plugin is already in rules_scala :D https://github.com/bazelbuild/rules_scala/blob/master/third_party/plugin/src/main/io/bazel/rulesscala/dependencyanalyzer/DependencyAnalyzer.scala I'm still curious if the original plugin detected extra jars if...

Okay, I quickly hacked together a really rough implementation of `unused_deps` in #438 that is bundled with `strict_java_deps`. I would like to get people's feedback before going any further.

We've been using both together at Lucid for a few months with Annex and it has worked quite well. Zinc's analysis is used to determine unneeded and unused deps and...

There would need to be one index per `scala_` target, right? The index would contain the mapping of class to label for the transitive closure of the current target? If...

Sorry for the lack of updates. I have an afternoon to sit down and look at this today. I'll let you know what I find.

I was able to work around this by replacing everywhere I was using `@org_scala_lang_modules_scala_xml_2_11//jar` and `@org_scala_lang_modules_scala_parser_combinators_2_11//jar` with `@scala//:scala-xml` and `@scala//:scala-parser-combinators`. We can now build everything with `--strict_java_deps=ERROR`. :tada: Thanks for...