Recently I uploaded the new Test::Most. From the end user perspective, the only real change (aside from being a touch easier to install), is that if you ask it to die when a test fails, it no longer just dies. Instead, it throws a Test::Most::Exception. The vast majority of people will never, ever need this feature. However, our test suite gets a bit tricky at times and we do things like this:
foreach my $test (@tests) {
my $tests_finished = eval { $test->run };
if ( my $error = $@ ) {
report_error( $test, $error );
}
else {
...
}
}
Internally, the way we report failures depends very much upon whether or not the tests halted because Test::Most was told to halt on failures, or whether they really died. Now I can just do this in the &report_error sub:
if ( eval { $error->isa('Test::Most::Exception') } ) { ... }
And much grief is saved.
In other news, I've been asked to add timing data to Test::Aggregate and I've thought that prove's state mechanism should possibly be extended to capture aggregated state information. In other words, while I didn't really intend to, I'm writing yet another new test harness. You would think I've learned my lesson after writing the new Test::Harness (also by accident, I might add).
And yes, I've toyed with colored test output for it ...