use Test2 in run()

- Replaces cmp_deeply() with is().

- Replaces the diagnostic with what I hope is a better diagnostic aid; a table
  showing what was expected and how it differed from what we wanted.

- This update probably makes lancew#10 a moot point.

.travis.yml

@@ -9,6 +9,6 @@ perl:
   - "5.12"
   - "5.10"
   - "5.8"
-install: cpanm -nq Devel::Cover::Report::Coveralls Moo Test2::Suite Test::Deep::NoTest Types::Standard
+install: cpanm -nq Devel::Cover::Report::Coveralls Moo Test2::Suite Types::Standard
 script: PERL5OPT=-MDevel::Cover prove -lv
 after_success: cover -report coveralls

lib/Scientist.pm

@@ -5,8 +5,9 @@ use warnings;
 # VERSION
 
 use Moo;
-use Test::Deep::NoTest qw/deep_diag cmp_details/;
-use Time::HiRes 'time';
+use Test2::API qw/intercept/;
+use Test2::Tools::Compare qw/is/;
+use Time::HiRes qw/time/;
 use Types::Standard qw/Bool Str CodeRef HashRef/;
 
 # ABSTRACT: Perl module inspired by https://github.com/github/scientist
@@ -90,15 +91,23 @@ sub run {
         $run_control->();
     }
 
-    my ($ok, $stack)    = cmp_details( \@candidate, \@control );
+    # Capture the events generated by is().
+    my $events = intercept {
+        is(\@candidate, \@control);
+    };
+
+    my ($ok_ev, undef, $why_ev) = @$events;
+
+    my $ok = $ok_ev->pass;
     $result{matched}    = $ok ? 1 : 0;
     $result{mismatched} = $ok ? 0 : 1;
 
     $result{observation}{candidate} = $wantarray ? @candidate : $candidate[0];
     $result{observation}{control}   = $wantarray ? @control : $control[0];
 
     if ($result{mismatched}){
-        $result{observation}{diagnostic} = deep_diag($stack);
+        # $why_ev has a table displaying why the test failed.
+        $result{observation}{diagnostic} = $why_ev->message;
     }
 
     $self->result( \%result );

lib/Scientist.pod

@@ -99,7 +99,10 @@ Result contains the result of the experiment after it is run.
 
 Will contain data only AFTER C<-E<gt>run()> has been called.
 
-Observation is included in the result, this inclides macthed/mismatched. If mismatched present will also have diagnostic present.
+Observation is included in the result, this includes matched/mismatched. If
+there was a mismatch, then a diagnostic will be made available. This will
+display what was expected (from the control) and what we got instead (from the
+candidate).
 
 =item run()
 

t/scientist.t

@@ -22,9 +22,17 @@ is $experiment->result->{'mismatched'}, 1,
 is $experiment->result->{observation}{candidate}, 20, 'Observation Candidate data correct';
 is $experiment->result->{observation}{control},   10, 'Observation Control data correct';
 
-like $experiment->result->{observation}{diagnostic},
-     qr/got : '20'/,
-     'Observation diagnostic correct';
+my $expected_diag = q{
+	+------+-----+----+-------+
+	| PATH | GOT | OP | CHECK |
+	+------+-----+----+-------+
+	| [0]  | 20  | eq | 10    |
+	+------+-----+----+-------+};
+$expected_diag =~ s/^\s+//mg;
+
+is $experiment->result->{observation}{diagnostic},
+	$expected_diag,
+	'Observation diagnostic correct';
 
 $experiment->use( \&old_code );
 $experiment->try( \&old_code );