> OpenVMS V7.3-2 on an Alpha, Q D E S I G N (7.10.G1)
> Copyright 1998 COGNOS INC. (ALPHA)
> Licensed PH-AXP-DEVELOPMENT
>
> Have a detail file in a screen that occurs 5 times
> FILE XYZ-LINES DETAIL OCCURS 5 TIMES CACHE 255 NOITEMS
>
> In testing a screen, I just updated 9 detail records and received this message which I have never
> seen before:
>
> *d* The occurrence count exceeds the defined occurrence for this file.
>
> I looked at the 9 records in QUIZ and everything appears to be normal. When I do a QUICK (F)ind,
> all 9 records show up normally.
>
> But surely a message like this would alarm a user. Anybody have a clue?
Hmmm...
A *d* error usually means that the screen in question did not compile properly or has not been written properly - it's a design error.Ã, Do you by any chance have a custom-coded update procedure or designer procedure that's trying to loop through the cluster doing something "special"? My guess is that a piece of procedural code somewhere is setting the occurrence variable to a value higher than the number of items in the cluster.Ã, If you're not sure what's causing the error, try compiling the screen in debug mode, and step through the code a line at a time in the debugger until you get the error, or alternatively set a watchpoint on occurrence to see when it changes and what's changing it.Ã, That may at least point you to the part of the code that's "suspect".
If you want to prove it's the screen, try writing a very simple default screen with a cluster for the same file from scratch.Ã, If you get the same error, then log it with Cognos Telesupport, otherwise it points to a problem somewhere in your code.
Sorry to be vague, but these kinds of errors often take some time and effort to pin down.
Good luck!
MF
MF is write. According to Bob Deskin of Cognos a *d* message indicates a design/er error. I.e. a problem with the code logic. I agree that using debug is probably the most productive way to go. That nad get another pair of eyes to go over the source code.
Received info from other sources that this was an old bug in QDESIGN/QUICK that was corrected in a later version. But since client refuses to upgrade, we'll just avoid using CACHE with the old version.