The Zdatr 2.0 engine is a DATR implementation in ANSI C for UNIX and Windows 95. The recommended compiler for the source code distribution is gcc. Zdatr was originally designed in 1995 to provide convenient high-speed batch or interactive standard DATR query processing with large DATR theories, convenient I/O and system IPC interfaces to CGI and other processes.
The requirements and design specifications and the implementation architecture are based on almost a decade experience in implementing DATR for teaching and project applications in Bielefeld in a number of different styles: in Scheme (ddatr), for teaching purposes, Prolog (node), for generating the large VERBMOBIL project lexicon, and with GNU tools, mainly gawk (mud), for driving a CGI interface on the web.
The Zdatr 2.0 system is split into two independent modules, a tokenizer (zdatrtok) for preprocessing of a theory and the inference engine (zdatrinf), which permits queries to be processed via manual input, query declaration files or query list files (see 2.6 for details). An interface is also provided for a number of standard invocations of Zdatr 2.0.
The Zdatr 2.0 library consists of RFC 2.0 compliant node definitions and from the user's point of view, they can all be regarded as builtins. For the more technically minded DATR user, the library functions are implemented in two different ways:
The version changes history is documented separately in the Zdatr 2.0 distribution. The wish list for modifications in the next minor release includes removal of a known bug (see Section 5) and preparation of documentation in texinfo format in addition to the DVI, PS, HTML and LaTeX files provided with the Zdatr 2.0 distribution.