Browse Books

Reviewer: Raphael M. Malyankar

Domain-specific languages (DSLs) deal with focused problems, usually within the context of a larger system or software tool. They generally focus on narrow, well-defined aspects of a software system, such as configuration, or on modeling a specific class of problems in a well-defined domain. Whether called by name or not, DSLs turn up in many software development situations; any programmer who has used some flavor of regular expression languages in general-purpose languages, such as in Java, C#, or Perl, or the configuration files for Struts or Hibernate, has encountered one. Software architects, developers, and even engineers in other disciplines who use design or simulation software will encounter modeling or configuration tools that clearly need a DSL of one kind or another. Architects, designers, and developers of software systems of even moderate size and complexity will often need to invent and implement such a language for some aspect of the software system, extend an existing language, or improve support for it in the system. This book deals with many of the questions that will arise during the definition, design, and development of such DSLs. The overall structure of this book is what Fowler calls a "duplex book": a relatively short narrative text communicates the concepts and basic ideas, and is intended to be read in full; the reference material should be read as needed. The first quarter of this book (Part 1) is a practitioner's textbook: a mixture of fundamental concepts in computer science (CS) and compilers; practical advice about when to apply the concepts; explanations of the trickier or less obvious points; and guidelines and tips for implementation. This section includes tips on when to use a DSL, discussions of the DSL life cycle, language idioms, tips for language design and construction, models of computation, constraints and tradeoffs in using DSLs, and short discussions of the basic concepts of programming language design and related issues (for example, interpreted versus compiled languages and compiler implementation). The final chapter in Part 1 discusses language workbenches. But, as the book makes clear, the state of the art in language workbenches is such that language designers can only draw one conclusion from this chapter: workbenches have potential, and it might be worth looking for one, but don't depend on finding one suited to particular circumstances. The first chapter in Part 2 contains very brief overviews of a few popular DSLs, including cascading style sheets (CSS) and Hibernate query language (HQL). This part should be read with Part 1. The rest of the book is a handbook for software architects, designers, and developers. It is a collection of conceptual building blocks that range from basic programming language concepts, such as lexical analysis and recursive descent parsing, to more advanced ideas, such as production rules. Each chapter in this reference portion of the book (Parts 2 through 6) has the same general structure: describe a DSL design/implementation building block, and then discuss its use, functioning, and implementation. It includes code fragments in Java, C#, and Ruby. Though much of the book discusses fundamental ideas in programming languages and compiler theory and implementation, this is not a textbook for learning programming language theory or compiler design. Readers need to have a basic knowledge of CS theory (specifically, programming languages and automata), such as what might be gained from an undergraduate CS course; undergraduate-level (at least) coursework in compilers would also be useful. Practitioners familiar with compiler theory will have already encountered many of the techniques described here; however, they should still find it useful to have the relevant software patterns collected in one volume for handy reference. For those without backgrounds in compilers or programming language design, there is still useful material in the first part. Motivated software architects and developers who have the appropriate CS coursework or experience may find the significant time and effort required to learn the necessary background material in compilers and programming worthwhile, especially if they are designing and developing new, complex software systems, which are likely to have features or requirements where a DSL would be an appropriate part of the solution. Online Computing Reviews Service