Explicitly exit with the correct status

Why:

* When running tests on a directory, the exit status was 0, regardless of the
  actual exit status of the runner
* CI builds depend on status codes for each step in the build process to
  determine whether the build passed or failed. Since `exit()` is being used
  here, test suites that experience runtime errors will still exit with a
  0 status code.

This change addresses the need by:

* Exiting with 1 when there is a `RuntimeError`
* After the rake task runs, exit with the exit status of the runner
* Creates a method to assert output when the underlying process failed, instead
  of assuming that it will always be successful

lib/error_tests/error_test.rb

@@ -0,0 +1,7 @@
+require_relative "../../test/test_helper.rb"
+
+class ErrorTest < MTest
+  def test_purposeful_error
+    raise RuntimeError
+  end
+end

lib/m/parser.rb

@@ -31,9 +31,9 @@ def parse
           begin
             Rake::Task['m_custom'].invoke
           rescue RuntimeError
-            exit
+            exit(1)
           ensure
-            exit
+            exit($?.exitstatus)
           end
         else
           return testable

test/everything_test.rb

@@ -27,7 +27,7 @@ def test_running_tests_within_a_subdirectory
 
   def test_running_tests_with_failures_within_a_subdirectory
     output = m('examples/subdir_with_failures')
-    assert_output(/1 tests, 1 assertions, 1 failures/, output)
+    assert_output_for_failed_execution(/1 tests, 1 assertions, 1 failures/, output)
   end
 
   def test_blank_file_is_quieter

test/exit_codes_test.rb

@@ -0,0 +1,33 @@
+require "test_helper"
+
+class ExitCodesTest < MTest
+  def test_failing_test_returns_1
+    m("examples/subdir_with_failures/a_test")
+    refute $?.success?, "expected exit code to be 1 but it was #{$?.exitstatus}"
+  end
+
+  def test_test_with_error_returns_1
+    m("../lib/error_tests/error_test")
+    refute $?.success?, "expected exit code to be 1 but it was #{$?.exitstatus}"
+  end
+
+  def test_dir_with_failure_returns_1
+    m("examples/subdir_with_failures")
+    refute $?.success?, "expected exit code to be 1 but it was #{$?.exitstatus}"
+  end
+
+  def test_dir_with_error_returns_1
+    m("../lib/error_tests")
+    refute $?.success?, "expected exit code to be 1 but it was #{$?.exitstatus}"
+  end
+
+  def test_without_errors_or_failures_returns_0
+    m("examples/subdir/a_test")
+    assert $?.success?, "expected exit code to be 0 but it was #{$?.exitstatus}"
+  end
+
+  def test_dir_without_errors_or_failures_returns_0
+    m("examples/subdir")
+    assert $?.success?, "expected exit code to be 0 but it was #{$?.exitstatus}"
+  end
+end

test/test_helper.rb

@@ -16,6 +16,11 @@ def assert_output(regexp, output)
     assert $?.success?, "Execution failed, output:\n\n#{output}"
     assert_match regexp, output
   end
+
+  def assert_output_for_failed_execution(regexp, output)
+    refute $?.success?, "Execution did not fail, but it should"
+    assert_match regexp, output
+  end
 end
 
 require 'm'