NAME Test::Future::IO - unit testing on Future::IO SYNOPSIS use Test::More; use Test::Future::IO; my $controller = Test::Future::IO->controller; { $controller->expect_syswrite( "Hello, world\n" ); $controller->expect_sysread( 256 ) ->returns( "A string\n"); code_under_test(); $controller->check_and_clear( 'code under test did correct IO' ); } DESCRIPTION This package provides a means to apply unit testing around code which uses Future::IO. It operates in an "expect-and-check" style of mocking, requiring the test script to declare upfront what methods are expected to be called, and what values they return. EXPECTATIONS Each of the actual Future::IO methods has a corresponding expectation method on the controller object, whose name is prefixed with expect_. A single call to one of these methods by the unit test script represents a single call to a Future::IO method that the code under test is expected to make. The arguments to the expectation method should match those given by the code under test. Each expectation method returns an object which has additional methods to control the behaviour of that invocation. $exp = $controller->expect_sleep( $secs ) $exp = $controller->expect_sysread( $len ) $exp = $controller->expect_syswrite( $bytes ) Note that the sysread and syswrite expectations currently ignore the filehandle argument. The returned expectation object allows the test script to specify what such an invocation should return. $exp->returns( @result ) METHODS controller $controller = Test::Future::IO->controller; Returns the control object, on which the various expect_* methods and check_and_clear can be invoked. check_and_clear $controller->check_and_clear( $name ); Checks that by now, every expected method has been called, and emits a new test output line via Test::Builder. Regardless, the expectations are also cleared out ready for the start of the next test. TODO * Some Test::Deep integration to be less fragile on things like sysread length. * Configurable matching on filehandles. Provision of a mock filehandle object to assist unit tests. * Ability for expectations to fail. AUTHOR Paul Evans