* jmh normalizes jar timestamps
build_is_identical e2e test spits out differences in md5

* change jar creator to default to normalizing timestamps

build_is_identical stabilization

)

* Output statsfile with line 0 = compilation time

* Actually specify that the statsfile is an output :)

* Use Java nio utilities to write statsfile; adjust to property file inspired syntax

* Add a test for the statsfile output

…lbuild#409)

When executing non-locally, the wrong version of Java would be used,
referencing a path that didn't exist. This change adds the
_java_runtime attribute to have to locate the java runtime on the
machine where the Scala code is being executed, and selects the
correct configuration for the host_javabase.

As a minor cleanup, this also removes the now no-longer-necessary
_java attribute.

TESTED=Ran unit tests and also ran a custom Scala unit test through:
previously, the Scala unit test failed with an incorrect java path
error, now it passes.

Change-Id: I9c2829bded04ec2d9d014094bd9e72cd984fe6e6

Resolves: bazelbuild#412

The IntelliJ plugin assumes that source_jar and source_jars are both
present on the jars struct. This sets those to empty values, so the
IntelliJ plugin does not fail.

Without it, builds fail on Windows as it uses semi-colon instead.

…with specs2 v4

* use transitive_proto_path_flags if exist

* Update scala_proto.bzl

…rap-code

Compile reporter using very strict settings

Add default javac opts when compiling java srcs

…_path

adjust transitive_proto_path_flags skylark name to transitive_proto_path

This enables targets that are files to be used with scala_import. For example
if you have a jar checked into your repository, you can use that with
scala_import.