The Next 700 Programming Language Manuals

The first programming manual [WWG51] was produced in 1951 by Wilkes, Wheeler and Gill. It was a carefully written scientific report, providing an account of distilled experience for the tiny global community of programmers [CK11]. Now 60 years later, we wonder whether there is still a need to write traditional programming manuals at all [SH01]? If there is, can we crowd-source the production by enabling early adopters to draft the documentation, and subsequently leaving the text to evolve online in a Wikipedia style [KCP07]? Documentation for programming languages is particularly important, in relation to other software documentation, e.g. that of an application or a user library. It is difficult for a programmer to engage in exploratory development without at least a basic understanding of the language-specific concepts introduced in the manual. In this project, we will investigate issues like: What does a useful manual need to tell us? How long is it? How formally should it be written? Does the documentation quality affect the popularity of the language? We know that once a programming language acquires a critical mass of programmers, the number of published textbooks increases dramatically. However, are there simple steps that a language inventor can take in order to make this popularization more likely? In this PhD research, we propose to examine how programming language documentation has changed over time. We will investigate techniques that are effective, and those that are not. Our goal is to assess how programming language manuals could evolve in the future to be more relevant for the next generation of programmers. 2 Investigator Profile