Scott McWhirter wrote:
> This would be easy to do in Test::Builder within &Test::Builder::ok by
> making it call $Test->BAIL_OUT();
>
> Then at least you would get it across all Test::* modules.
yeah, that's what I was implying in my followup message.
the attached patch against 0.70 seems to do what I had in mind.
--Geoff
--- lib/Test/Builder.pm 2007-09-07 15:33:04.000000000 -0400
+++ lib/Test/Builder.pm! 2007-09-07 15:32:30.000000000 -0400
@@ -196,6 +196,8 @@
$self->{No_Header} = 0;
$self->{No_Ending} = 0;
+ $self->{Bail_Out_on_Failure} = 0;
+
$self->_dup_stdhandles unless $^C;
return undef;
@@ -457,6 +459,10 @@
}
}
+ if (!$test && $self->{Bail_Out_on_Failure}) {
+ $self->BAIL_OUT('failure detected and bail_out_on_failure() set');
+ }
+
return $test ? 1 : 0;
}
@@ -1130,9 +1136,17 @@
If set to true, no "1..N" header will be printed.
+=item B<bail_out_on_failure>
+
+ $Test->bail_out_on_failure($bail_out);
+
+Normally, Test::Builder continues right along on when a test fails.
+By setting this to true a failure will call BAIL_OUT() and all
+activity will be wrapped up as appropriate.
+
=cut
-foreach my $attribute (qw(No_Header No_Ending No_Diag)) {
+foreach my $attribute (qw(No_Header No_Ending No_Diag Bail_Out_on_Failure)) {
my $method = lc $attribute;
my $code = sub {
