The Use of Literate Program Development Tools

1. Introduction
2. The Philosophy of SWR Pascal

   	Background
   	Why SWR Pascal?

3. Experiences of Using SWR Pascal
4. Future Trends and Other Tools
5. References
6. The SDG

	Problems of separation of source code from documentation and the recognised benefits of self-documenting programs;
	The need to perform good program design;
	The maintenance problem and its exacerbation by 'hacking existing code' and thinking at the terminal;
	Programs are more easily comprehensible if they are legible and pleasing to look at;
	Literate programming, as espoused by Knuth [KNU84], was recognised as being a worthy technique [PAR87].

	Need for a straightforward, easy to use, design method: The SDG liked the idea of Step Wise Refinement, as espoused by Wirth [WIR71]; Also recognised Jackson's criticisms of the technique "design by direct onslaught" [JAC75b]; Devised a method of SWR using analysis of sample data to derive a program structure which can then be translated into SWR Pascal.
	SWR Pascal uses structured techniques to enforce design and thus to eliminate 'thinking at the terminal';
	It marries code to design forcing the programmer to consider the program in light of the design at all times. There is only a single source document from which both the technical documentation and executable code are derived, thus, code and documentation are inextricably linked and updated together.

	Used successfully since 1985;
	Brings students face to face with design at an early stage by use of ready-designed skeleton programs;
	Discourages thinking in terms of pure code as it interweaves code with documentation/design. This leads to students having to examine programs in the light of their design;
	The group has insisted upon presentation of sample data and structure diagrams in coursework assignments. Actual code counts for relatively little;
	SWR Pascal discourages the student who is inclined to go straight to the coding phase. Because SWR Pascal forces the student to design his program and to think in terms of program structure and the structure of sample data, it has been a useful stumbling block to the unrepentant 'hacker'. The program code is broken up into small sections and is interspersed with design statements. This makes direct 'hacking' of the code very difficult;
	SWRPAS possesses a monitoring facility which enables the staff members involved to review student performance in terms of the number of attempts to compile and run a program. Coupled with the continual assessment of the students' initial progress [SDG91] this facilitates a fast system of feed-back and quickly identifies early problem cases;
	The tool is used by its authors for all software development carried out in the VAX environment. Early versions of SWR Pascal were used to provide the development platform for later versions;
	It works! Experience has shown that the programmer who spends a few hours sketching the sample data, producing a program structure and from that deriving SWR Pascal code will be more likely to achieve first time compilation and execution.

	It is intended to implement SWR Pascal in a PC-based environment, although this is expected to be superseded quickly by-
	A folding editor. The group is currently developing a folding editor environment as a replacement for the Borland Turbo Pascal IDE. This will combine all of the features of SWR Pascal (design statements being replaced by folds) with the advantages of a more user-friendly development environment.
	There is a stand-alone structure diagrammer for the VAX which, it is anticipated, will migrate to a PC.
	Members of the group have written a JSP pre-processor to support the group's teaching of Jackson Structured Programming. This is a COBOL-based tool for the VAX. There are no plans to implement this in MS-DOS for the use of VAX/VMS gives the student valuable experience of a mainframe environment.

The Software Development Group in the School of Computing and Mathematical Sciences is a research and development grouping. Its members are responsible for the teaching of program design and software engineering to the students in the School. The principal research area is commercial software maintenance to which the concepts of literate programming are particularly applicable. Other areas of interest include software estimation, quality and the provision of CASE tools.