Section 5.2.6.1.  OOSTAR

The OOSTAR approach (Chang & Bourne, 1998) developed and tested an Objective-C (Pinson & Richard, 1991) toolset for STAR representation. The developers selected Objective-C over the more common C++ language because of its perceived advantages of (i) loose data typing, (ii) run-time type checking and (iii) message-passing ability. These were all seen to aid flexible software design, an asset for applications built on the very flexible structure of the STAR syntax itself.

Chang & Bourne constructed a set of classes built on the basic objects item and value. Their model builds upwards through a hierarchy of classes describing relationships between objects. The ItemAssoc class contains data members corresponding to the data item itself, its included value or set of values, and the data name that is used to reference the item; the class also has pointers to previous and next elements to allow iteration. The StarAssoc class contains methods for manipulating STAR objects through setting and retrieval of data names and assignment of values. DataBlock is the class containing all the lower-order items and associations, and one or more such DataBlock classes comprise the StarFile class.

The OOSTAR approach also provides Dictionary and DictionaryElement classes so that STAR applications that do describe element attributes in external dictionary files can make use of such information.

Some sample applications were built with the class libraries of the OOSTAR toolset, and they are described in Chang & Bourne (1998). They include: simple converters of STAR data files to HTML pages with hyperlinks to associated dictionary entries; a query tool for retrieving the data values for a specific item specified by name within a specific data block; and a query mechanism that retrieves a set of items from a STAR loop structure and represents their values in an array, i.e. flattening if necessary any nested loops into a purely tabular presentation. These query tools are different from Star_Base primarily in that they do not return the context in which the data are found in the original file – their interpretation depends on a detailed a priori understanding of the target file structure.

This approach is rather different from the canonical description of STAR given in this volume, and was never developed into full-blown applications (in part because the developers recognized that the Java language would provide a preferable platform for further development). Nevertheless, it provides interesting ideas for the developer considering building object-oriented applications for STAR Files.