Tycoon/Fibonacci Add-On Experiment(ID:6916/tyc003)


Related languages
Fibonacci => Tycoon/Fibonacci Add-On Experiment   Incorporated into
Modula-3 => Tycoon/Fibonacci Add-On Experiment   Written using
TYCOON => Tycoon/Fibonacci Add-On Experiment   Incorporated some features of
References:
  • Davide Berveglieri, Florian Matthes, Gerald Schröder "The Tycoon/Fibonacci Add-On Experiment" view details Abstract: The FIDE-1 research has established a common framework for the analysis and design of type systems in the area of database and persistent programming languages [A*89]. Based on this common framework and their particular past experiences, the groups at Pisa University and Hamburg University designed next-generation database programming languages, Fibonacci [ABGO93] and Tycoon [MaSc92a], respectively.
    Within FIDE-2, both groups are using FIDE technology (object stores, PHOLAMs, type checking algorithms) to build language processors for their new languages. Some of this implementation work is covered by funding outside FIDE-2. The Hamburg team allocates a significant amount of labour and technology to a high-quality Tycoon system implementation while the focus of the Pisa team is on an initial prototypical implementation that enables rapid experiments to investigate data modeling issues.

    The Tycoon system is based on a small, portable core runtime system written in C which subsumes a low-level object store interface and an interpretative evaluator. All other Tycoon system components (parser, type checker, code generator, optimizer, system browser, library manager, ...) are written in the Tycoon language. In February 1994, the bootstrap of the Tycoon system has been completed, i.e. the system is able to compile itself and each further improvement of the compilation technology immediately improves the performance of the Tycoon system. As of today (February 1994), the Tycoon system implements the full language as defined in [MaSc92a] except dynamic type checking at run-time.

    A subset of the Fibonacci language (without parametric polymorphism, module mechanisms, separate compilation, triggers) has been implemented in Modula-3. The Fibonacci compiler is written im Modula-3 and generates a PHOLAM code that is executed by an interpreter also written in Modula-3. The PHOLAM code contains high-level bulk data operators (select, project, join, ...) to facilitate optimizations in a future project phase. These bulk data operations are implemented by Modula-3 subroutines.
    External link: Online copy