Pierre Boullier, a researcher at INRIA in France, quietly built a system that would eventually become the backbone of many critical software tools without ever seeking the global fame enjoyed by his contemporaries. SYNTAX, the compiler-generation system he spearheaded, was designed to handle the complex task of creating lexical and syntactic analyzers for all kinds of context-free grammars, a feat that required decades of meticulous work. Unlike many open-source projects that gained immediate traction, SYNTAX remained a specialized tool within the French research community until 2007, when it finally became free software under the CeCILL license. This delay in public release meant that the system's true potential was realized only after it had already proven its worth in high-stakes operational environments, such as the development of the first tool-translator for the Ada programming language. The story of SYNTAX is not one of viral marketing or rapid adoption, but of steady, unyielding progress in the face of technical challenges that stumped other researchers.
The Ada Compiler Breakthrough
The first tool-translator for the Ada language, developed by Pierre Boullier and his team, stands as a testament to the power of SYNTAX in handling deterministic grammars. This project, which took place between 1978 and 1981, was a significant milestone in the history of programming language compilation. The team, including Knut Ripken, successfully built an Ada compiler following meta-compilation methods, a process that was both innovative and technically demanding. The resulting compiler was not just a theoretical exercise but a practical tool that demonstrated the capabilities of SYNTAX in real-world scenarios. The work was published in the Séminaires Langages et Traducteurs 1978-1981, pages 99-140, by INRIA in Rocquencourt, France. This achievement laid the groundwork for future applications of SYNTAX in the domain of compilation, showcasing its ability to handle complex grammatical structures with precision and efficiency.Parsing Beyond the Deterministic
SYNTAX's capabilities extended far beyond the deterministic grammars that had initially defined its reputation, venturing into the realm of non-deterministic parsing with the introduction of an Earley parser generator. This feature was particularly valuable for natural language processing, where the ambiguity of human language often defies simple grammatical rules. The SxLFG parser, whose first version was described in a dedicated paper, exemplified the system's versatility in handling context-sensitive formalisms such as TAG, RCG, and LFG. These formalisms, which rely on context-free grammars extended by attribute evaluation, allowed for more nuanced and accurate parsing of natural language. The inclusion of these advanced features in version 6.0 beta of SYNTAX marked a significant evolution in the system's capabilities, enabling it to tackle problems that were previously beyond the reach of traditional parsing tools.