Web Enablement: Challenges and Solutions Revision 1

In January 2003 I discovered an error in my paper presented at the 11th Natural Conference, Web Enablement: Challenges and Solutions. Actually the error was introduced after the original paper (HTML document) was submitted to Jim Wisdom, and thus resided in the copy on the University of Arkansas web site. This document descibes the cause for the error and the correction, which has been incorporated into the document -- thus the revision date.

The error had to do with the topic of Browsing ADABAS Data for MU Type Descriptors. It specifically involved the value used for #START-ISN with the READ ... STARTING WITH ISN which is coded within the HISTOGRAM loop. The paper indicates that this value is reset to zero after the first time through the HISTOGRAM loop since we only need this value to restart reading records for the first descriptor value following a page forward or backward request. Subsequently we want all records and a zero STARTING WITH ISN provides all records -- even when reading descending. I learned that it worked this way during the preparation of this paper. Previously I had coded (and what is running in production) the assignment of #START-ISN based upon the read direction as follows: 
    IF *COUNTER(HKEYW.) GT 1                                     
      /* Only restrict the records in the READ for the first key 
      IF #DIRECTION = 'A'                                   
        RESET #START-ISN                                         
      ELSE                                                       
        #START-ISN := 4294967295 /* Highest possible ISN         
      END-IF                                                     
    END-IF
When I learned that a read descending starting with ISN value of zero would provide all records, I assumed it was much preferred over setting the value to the highest possible ISN. What I did not realize is that this introduced a bug in the subsequent code that checks for reading the same record two more times for the same descriptor value. This is where the correction has been made to the paper. The original code provided for this purpose is: 
        IF NOT HKEYWV.#MENU-KEYWORD = DBV.MENU-KEYWORD(*)
            OR /* Ensure we don't cycle around           
              (#DIRECTION = 'A'                          
                AND *ISN(RKEYW.) LE #START-ISN)          
            OR                                           
              (#DIRECTION = 'D'                          
                AND *ISN(RKEYW.) GE #START-ISN)          
          ESCAPE BOTTOM                                  
        END-IF                                           
        #START-ISN := *ISN(RKEYW.)
This worked fine when the #START-ISN value was the actual record we wanted to start with or high values, but it does not work with a zero value. The corrected code I have chosen to use follows (although another alternative would be to use the high value ISN when reading descending): 
        IF NOT HKEYWV.#MENU-KEYWORD = DBV.MENU-KEYWORD(*)
            OR /* Ensure we don't cycle around           
              (#DIRECTION = 'A'                          
                AND *ISN(RKEYW.) LE #START-ISN)          
            OR                                           
              (#DIRECTION = 'D'
                AND #START-ISN NE 0  /* < -- added statement                          
                AND *ISN(RKEYW.) GE #START-ISN)          
          ESCAPE BOTTOM                                  
        END-IF                                           
        #START-ISN := *ISN(RKEYW.)
My sincere apologies to anyone that has been mis-lead by this error. It certainly caused some difficulties here at the U of A.