我正在修复一个大型测试脚本(> 1000行),它使用一些实用工具方法(也包括1000行)对各种初始数据设置执行重复测试。这有助于整合代码。但是,当测试失败时,它会报告实用程序方法内部的行号,这使得很难跟踪哪个测试失败。
当测试失败时,是否可以将Test::Most
配置为提供堆栈跟踪而不是单个行号?
#!/usr/bin/env perl
use strict;
use warnings;
use autodie;
use Test::Most tests => 3;
ok(1, 'first test');
note "The following includes a failed test, but a stack trace would be more helpful";
helper_sub_with_test(); # Line 13
ok(1, 'third test');
sub helper_sub_with_test {
ok(0, "second test"); # Line 17
}
输出:
$ perl scratch.pl
1..3
ok 1 - first test
# The following includes a failed test, but a stack trace would be more helpful
not ok 2 - second test
# Failed test 'second test'
# at scratch.pl line 17.
ok 3 - third test
# Looks like you failed 1 test of 3.
正如您所看到的,如果失败的测试报告了对实用程序方法进行多次调用时第17行和第13行,将会很有帮助。
答案 0 :(得分:4)
我不相信Test :: More基础设施提供了这样的野兽,但你真的需要一个堆栈跟踪吗?只要您为测试提供描述性名称,报告第13行就足够了。
要报告第13行而不是第17行,只需将以下内容添加到您的子文件中:
local $Test::Builder::Level = $Test::Builder::Level + 1;
更长的例子:
#!/usr/bin/env perl
use strict;
use warnings;
use autodie;
use Test::Most tests => 3;
ok(1, 'first test');
note "The following includes a failed test, but a stack trace would be more helpful";
helper_sub_with_test(); # Line 13
ok(1, 'third test');
sub helper_sub_with_test {
local $Test::Builder::Level = $Test::Builder::Level + 1;
ok(0, sprintf "second test (look at line %d)", __LINE__); # Line 18
}
答案 1 :(得分:3)
获得所需内容的快速而肮脏的方法是在Test :: Builder :: ok周围放置一个包装器。这就是Test :: Most的运作方式。
使用Aspect可以减少黑客攻击。
use Carp qw(longmess);
use Test::Most;
use Aspect;
after {
# For some reason, the return value is not being captured by Aspect
my $last_test = ($_->self->summary)[-1];
print longmess if !$last_test;
} call "Test::Builder::ok";
sub foo { ok(0) }
foo();
pass;
done_testing;