It offers little in the way of defining classes of input which we believe to have similar errordetecting ability. In fact, the valid input space on its own is suitable only for deriving a suite of random tests, each a member of the valid input space. Nevertheless, the valid input space is a useful template to define and has an important role to play in the framework. As mentioned at the end of chapter 3, the valid input space of an operation must be the source of all specification-based tests for the operation. This means that any test is an element of the valid input space. It also means that any test template must be a subset of the valid input space. So, we can define a Z type for test templates for a certain operation, Op: TT Op == p VIS Op Note the subscripted use of the operation name. This is a practice we will adopt for the remainder of the thesis. This definition defines TT Op to be the type of all test templates for Op. Schemas vs sets: The significance of bindings It has already been noted that templates describe sets of test data and that Z schemas are used to define templates. It may seem strange not to use sets to define templates. As mentioned in section 4.1, defining test data for an operation involves assigning values to the input components (both state and parameter) of the operation, that is, defining a binding between input component identifiers and values. Thus, a template intuitively defines a set of bindings, which is exactly what a Z schema defines.

Download PDF

Back