This paper discusses the need for extensible database management systems (DBMSs), illustrates the benefits of using a functional data model and data language as the semantic interface to extensible DBMSs, and provides a good overview of one such system, GENESIS. The authors cover the implementation of GENESIS in detail at a conceptual level and informally define GDM, the data model upon which GENESIS is based, and GDL, the data language they use.
In the authors’ implementation, GDM productions are stream rewrite rules rather than mathematical functions and GDL computations are expressed as streams of tokens. The novelty of this approach is that it makes delimiters explicit in token streams. The authors present this key feature as the basis for a simple and efficient standardization of DBMS algorithms, which in turn promotes extensibility and module plug-compatibility. The work’s impact on the building of a non-1NF query processor is of secondary interest.
I found the paper clear and well organized. The references seem well researched and the pioneering papers upon which the work is based are noted in the text. While the authors assume a good foundation in the topic area, the reader will not need to have followed this research closely in order to understand the paper or appreciate their arguments. Those interested in functional data models and languages, non-traditional database applications, or non-1NF database implementations will find the paper worth reading.