lberki

I did a bout of debugging to see how to fix the newly added tests, and I still don't know :( I looked at `UnixDigestHashAttributeNameTest.testGetNioPath_externalUtf8[SHA-256]`. I do know how it...

on Linux: ``` $ cat X.java public class X { public static void main(String[] args) throws Exception { System.err.println(System.getProperty("sun.jnu.encoding")); java.nio.file.Paths.get(new String("🌱")); } } $ javac X.java $ LANG=en_US.ISO8859-1 java X...

> But the test isn't calling `Paths.get(new String("🌱"))`, it's calling `Paths.get(unicodeToInternal(new String("🌱")))` which should be equivalent to `Paths.get(new String("\xf0\x9f\x8c\xb1"))` (UTF-8 encoding of "🌱"), whose characters are all

Oof. Do I understand correctly that the root cause is that Python/Java rules require the Python/Java toolchains which are registered in `@rules_python` / `@rules_java`, therefore, if we don't depend on...

That, I hope, but it doesn't help by itself because there is no way to tell Bazel "I don't need `@rules_java`, pinkie promise, please don't ever look at it", is...

cc @comius -- this discussion might have some bearing on the eventual approach to decoupling rule implementations from the Bazel binary.

The point of putting in the legwork to make sure that a C++ hello world doesn't depend on Java is that then we know we don't have unexpected dependencies and...

Yeah, I'm supportive of the idea of being able to say "to the best of knowledge, my builds don't need Java and please error out if it touches anything that...

@tjgq @bendavini @ahumesky I'd much, much rather no one use that built-in zipper tool. It was added in support of Android rules, which are in the process of being removed...

I addressed the review comments and although the presubmits haven't quite passed yet, they look good so far so I'll context switch away from this. PTAL!