Resolving conflicts with testing

Jesse writes:
ethanrowe:tickets/master/2239 leaks state in the spec. After the spec
is run, Puppet::Application is left in a :restart_requested state, and
several important behaviors, particularly Puppet::Transaction#evaluate
are disabled.

It's order dependent, so changing the mtime of spec files makes the
failures appear and disappear.

This spec file was generally pretty good about keeping state from
getting out, but one test was missing a stub for a dangerous call.

I wouldn't be surprised if this cleared up other errors in testing.

1 files changed, 5 insertions, 1 deletions

diff --git a/spec/unit/daemon.rb b/spec/unit/daemon.rb
index 1bf8f5692..e163f4067 100755
--- a/spec/unit/daemon.rb
+++ b/spec/unit/daemon.rb
@@ -279,6 +279,11 @@ describe Puppet::Daemon do
     end
 
     describe "when reexecing it self" do
+        before do
+            @daemon.stubs(:exec)
+            @daemon.stubs(:stop)
+        end
+
         it "should fail if no argv values are available" do
             @daemon.expects(:argv).returns nil
             lambda { @daemon.reexec }.should raise_error(Puppet::DevError)
@@ -288,7 +293,6 @@ describe Puppet::Daemon do
             @daemon.argv = %w{foo}
             @daemon.expects(:stop).with(:exit => false)
 
-            @daemon.stubs(:exec)
             @daemon.reexec
         end