http://qs321.pair.com?node_id=11100035


in reply to Bailing out from Test::More after 'some' failures

Quick'n dirty. Uses the fact, that Test::More::builder() is not overridden.

use strict; use warnings; use Test::More; #-- builder() does not exist in Test::More, just intercept call hierar +chy... sub Test::More::builder { my $limit = $Test::More::BAILOUT_LIMIT; my $tb = Test::Builder::Module::builder(@_); my @tests = $tb->summary; my $passed = grep { $_ } @tests; my $failed = @tests - $passed; if ($failed >= $limit) { $tb->BAIL_OUT("Limit of $limit errors reached (passed=$passed, fai +led=$failed)!"); } return $tb; } #-- here, we set the limit $Test::More::BAILOUT_LIMIT = 2; # or i.e. ... = $ENV{BAILOUT_LIMIT} ok( 1, "okay"); is( 1, 2, "oops"); ok( 0, "test $_/3 will fail..." ) for (1..3); done_testing;

Result:

ok 1 - okay not ok 2 - oops # Failed test 'oops' # at ./tm.pl line 29. # got: '1' # expected: '2' not ok 3 - test 1/3 will fail... # Failed test 'test 1/3 will fail...' # at ./tm.pl line 30. Bail out! Limit of 2 errors reached (passed=1, failed=2)!

Can be done properly by deriving your own class, but you get the idea. You need one more test than the limit, though. The interception happens before the next test case is executed. Limit = N demands N+1 tests, but I guess that is acceptable...

Edit: This implements an individual per-test-file-limit. If you want the limit to encompass the whole test session, you'll need to maintain the counter's persistency and life-cycle across all sessions (i.e. using Storable).