Library code conventions

Some notation used in the standard library differs from that found in the traditional languages like C. In particular, fixed-length indexed arrays are called ``vectors.'' The name ``array'' is used for variable-length indexed arrays which allow adding and removing elements from both ends and can expand or shrink at runtime depending on the number of elements they are holding. The indexing operation is denoted by ``!'' (which is an infix version of fetch), because the traditional bracket notation ``[]'' is used by the Cecil language to indicate parameterization of objects and types. For example, printing out an array element (or, more generally, a table element) can be done like this: (a!i).print;.

Creation of new anonymous instances of objects at runtime (which corresponds to using the new constructor for classes in languages such as C++), is typically achieved by new_ methods which encapsulate the object constructor expressions. The name of such a method usually starts with new_ followed by the name of the (template) object whose instance is being created; sometimes a suffix is used to distinguish between different versions. However, this is merely a convention. (The only restriction imposed by the language design is that no anonymous instances of abstract objects may be created at runtime, since these objects are used to define the interfaces and share common code between their subclasses.) Notable exceptions from this convention are methods that construct cons cells, pairs, triples, and quadruples, which don't start with new_ for convenience. Here are some examples:

- create an uninitialized m_vector method new_m_vector[T](size:int):m_vector[T]; - initialize the created m_vector with filler method new_m_vector[T](size:int, filler:T):m_vector[T]; - use the init closure to compute the initial values method new_m_vector_init[T](size:int, init:&(int):T):m_vector[T];

Many operations which may produce an error or a special condition (e.g., indexing an array) take an optional last argument which is the closure to be invoked in such case. Thus, a programmer may either ignore the possibility of an error (which would halt the program) or handle the error condition (e.g., return a certain object when the array index is out of bounds). The default value of the closure invokes the error method, which prints a message, offers a debugging prompt, and then terminates the program. For example:

- call error if the key is not found method fetch(t:table[`Key,`Value], key:Key):Value; - invoke if_absent if the key is not found method fetch(t:table[`Key,`Value], key:Key, if_absent:&():Value):Value;