Dynamic grammars and semantic analysis

TLDR: It is shown that dynamic grammars have the formal power of Turing machines and an experimental system which implements a non ambiguous \sl dynamic parser is sketched and applications of this system for the resolution of some semantic analysis problems are shown.

Abstract: We define a dynamic grammar as a device which may generate an unbounded set of context-free grammars, each grammar is produced, while parsing a source text, by the recognition of some construct. It is shown that dynamic grammars have the formal power of Turing machines. For a given source text, a dynamic grammar, when non ambiguous, may be seen as a sequence of usual context-free grammars specialized by this source text: an initial grammar is modified, little by little, while the program is parsed and is used to continue the parsing process. An experimental system which implements a non ambiguous \sl dynamic parser is sketched and applications of this system for the resolution of some semantic analysis problems are shown. Some of these examples are non-trivial (overloading resolution, derived types, polymorphism, \ldots) and indicate that this method may partly compete with other well-known techniques used in type-checking.