# file t/testml/encode.tml %TestML 1.0 Title = 'Tests for AcmeEncode'; Plan = 3; *text.apply_rot13 == *rot13; *text.apply_md5 == *md5; === Encode some poetry --- text There once was a fellow named Ingy, Combining languages twas his Thingy. --- rot13 Gurer bapr jnf n sryybj anzrq Vatl, Pbzovavat ynathntrf gjnf uvf Guvatl. --- md5: 7a1538ff9fc8edf8ea55d02d0b0658be === Encode a password --- text: soopersekrit --- md5: 64002c26dcc62c1d6d0f1cb908de1435
This TestML document defines 2 assertions, and defines 2 data blocks. The first block has 3 data points, but the second one has only 2. Therefore the rot13 assertion applies only to the first block, while the the md5 assertion applies to both. This results in a total of 3 tests, which is specified in the meta Plan statement in the document.
To run this test you would have a normal test file that looks like this:
use TestML::Runtime::TAP; TestML::Runtime::TAP->new( testml => 'testml/encode.tml', bridge => 't::Bridge', )->run();
or more simply:
use TestML -run, -testml => 'testml/encode.tml', -bridge => 't::Bridge';
The apply_* transform functions are defined in the bridge class that is specified outside this test (t/Bridge.pm).
TestML is a generic, programming language agnostic, meta language for writing unit tests. The idea is that you can use the same test files in multiple implementations of a given programming idea. Then you can be more certain that your application written in, say, Python matches your Perl implementation.
In a nutshell you write a bunch of data tests that have inputs and expected results. Using a simple syntax, you specify what functions the data must pass through to produce the expected results. You use a bridge class to write the data transform functions that pass the data through your application.
In Perl 5, TestML is the evolution of the Test::Base module. It has a superset of Test:Base's goals. The data markup syntax is currently exactly the same as Test::Base.