In functional test case, directly call ui operation and do validation is not easy to be maintenance, not easy to handle ui changes. Someone suggest to add a middle layer (DSL layer) to resolve it: changer will keep implementing the DSL, and Domain specialist (maybe Test Developer) use DSL to enrich the testcase. Do you use this is a good idea or not? Do you use this layer in your functional test?
It is a good idea, applied and found working well in practice. Makes test case development faster (industry experiences say 10x), improve test coverage, allow to apply different test frameworks. See e.g. industry case reported at http://www.thinkmind.org/index.php?view=article&articleid=valid_2011_5_20_40049