This document is helpful for people who want to implement a new FunLog++ code generator or who want to maintain the original FunLog++ system software. This document might be helpful to some minor extent for system programmers who want to integrate the FunLog++ interpreter into a new environment or who want to understand the architecture of the FunLog++ interpreter.

The FunLog++ interpreter was developed as part of the Pro.M.D. system. It cooperates with a frontend program for user interaction and a SQL database system in order to store permanent results.
 
 

interpreter

compiler syntactic analyser syntax tree semantic analyser semantic list code generator

code file

executer deductor  (reads SQL) result list effector   (writes SQL)

• The FunLog++ compiler reads a source file and generates a target code file.
• The FunLog++ executer reads one or more code files and stores or displays results according to the main message.

The FunLog++-compiler serves two alternative functions, generating target language code and generating source file listings.
 

Generating target language code

This compiler function needs three consecutive steps:

• The syntactic analyser reads a source file and builds up a syntactic tree.
• The semantic analyser converts the syntactic tree into a semantic list.
• The code generator convers the semantic list into the target code file.

• A FunLog++ source file with extension ".FUN".
• A Pro.M.D. source file with extension ".RUL".
• A Prolog subset source file with extension ".PL" (not yet).

• A Prolog code file with extension ".LPA" or ".QOF".
• A Java code file with extension ":JAVA".
• A Delphi Pascal code file with extension ".PAS".

The FunLog++ compiler can output source listing files for the three different types of input:

• A FunLog++ listing output as plain text "*.FLI".
• A Pro.M.D. listing output as plain text "*.RLI".
• A Prolog++ listing output as plain text "*.PLI" (not yet).

fun.exe root("lipid.fun"):layout("search").

name ::= root("descriptor.fun"):name.

The FunLog++ executer consists of two consecutive subprograms, the deductor and the effector:

• The FunLog++ deductor reads one or more code files and generates the result list according to the main message.
• The FunLog++ effector put texts in files, database fields or dialog windows according to the result list.

FunLog++ deductor

The FunLog++ deductor typically consists of three main parts:

• The runtime starter loads all standard libraries and executes the FunLog++ main message.
• The runtime engine provides for a set of elementary calculations (e.g. for platform functions)..
• The runtime library defines standard functions in classes like failure/1, known/1, semiknown/1, or unknown/1.

FunLog++ effector

The FunLog++-effector is not yet implemented. In future its task will be to execute those list of store- and display commands which was returned as the result from the deductor. The effector is responsible for correct multiuser access and transaction control.

The effector has to be run immediately after the deductor. Additionally to the list of store- and display commands the effector receives the list of all external read access from the deductor in order to re-check the proper input-output relation before results are stored.

The overall task of the executer is then to be performed in usually steps:

1. Call the FunLog++ deductor and obtain the read-list and the write-list
2. Lock all locations and check-read all data from the deductor's read list and write list.
3. If check-read differs from primary-read then unlock all locations of the read- and write-list and repeat from step 1.
4. If check-read is identical to primary-read then store all result text in the wanted locations and unlock all.
5. Executer is ready.