(no subject)

[Mike Palandri]

Hi yall,

I'm puzzled by some behavior I encountered in QTP 7.10 F4 OpenVMS.  Here's 
a simple example distilled from the longer program:

A file has four items and four records with the following values:

   Item0   Item1   Item2  Item3
   AAAAAA  111111    1     A
   AAAAAA  111111    2     B
   AAAAAA  111111    2     C
   AAAAAA  111111    3     D



This QTP reads the file and creates a subfile:

CAN CLE
ACCESS TESTREC

TEMP TI-ITEM      INT*02
TEMP TI-REC-CNT   INT*10

ITEM TI-ITEM      INITIAL 0
ITEM TI-REC-CNT   INITIAL 0

SORT ON ITEM0,  ITEM1, ITEM2, TI-REC-CNT

ITEM TI-REC-CNT = TI-REC-CNT + 1
ITEM TI-ITEM = ITEM2 AT TI-REC-CNT RESET AT ITEM1

SUBFILE TT1 KEEP INCLUDE &
    ITEM0, &
    ITEM1, &
    ITEM2, &
    TI-REC-CNT, 
&
    TI-ITEM



Here's what ends up in the subfile:

Item0   Item1   Item2   TI-REC-CNT   TI-ITEM
   AAAAAA  111111    1              1      0
   AAAAAA  111111    2              2      1
   AAAAAA  111111    2              3      1
   AAAAAA  111111    3              4      2



TI-REC-CNT provides a control break on every record.  The break is used to 
assign the value of ITEM2 to the temp TI-ITEM.  Because TI-REC-CNT changes 
on every record, I thought that this assignment would take place on every 
record.

However, on the third record, where the value of Item2 is the same as Item2 
in the second record, the break never occurs, and the assignment of Item2 
to TI-ITEM doesn't happen.  The break truly does not occur;  I ran a trace 
and found "at end" processing of Ti-Rec-Cnt  on the first, third, and 
fourth complexes, but not the second:

# NEW RECORD COMPLEX ->  1
**** PROCESSING AT END OF TI-REC-CNT 1
****   ITEM  TI-ITEM =   : AT TI-REC-CNT
****     BEFORE              0
****     AFTER               1


# NEW RECORD COMPLEX ->  3
**** PROCESSING AT END OF TI-REC-CNT 3
****   ITEM  TI-ITEM =   : AT TI-REC-CNT
****     BEFORE              1
****     AFTER               2


# NEW RECORD COMPLEX ->  4
**** PROCESSING AT END OF TI-REC-CNT 4
****   ITEM  TI-ITEM =   : AT TI-REC-CNT
****     BEFORE              2
****     AFTER               3


However, if an additional sort key is inserted on the sort statement 
between Item2 and Ti-Rec-Cnt, the "At end" processing for Ti-Rec-Cnt occurs 
on all four complexes, and the code works as (I) expected.


New sort statement:
SORT ON ITEM0,  ITEM1, ITEM2, ITEM3, TI-REC-CNT



New resulting subfile:
   Item0   Item1   Item2   TI-REC-CNT   TI-ITEM

   AAAAAA  111111    1              1      0
   AAAAAA  111111    2              2      1
   AAAAAA  111111    2              3      2
   AAAAAA  111111    3              4      2


It seems to me that the "AT" for the lowest level sort key should be 
processed the same regardless of the value of higher level sort keys, or 
reports with headings at multiple levels of control breaks would not work.

Anybody know what's going on?


Thanks,


Mike
.