John,
I was curious about this. A quick test on PAL 2.0.9.17 and I replicated your results. So, I decided to try on a bigger cube with a longer running view where I could create and track the Stargate views and found an issue that may be related. With AllRuleCalcStargateOptimization=T and ViewConsolidationOptimization=T, and a view of all C: level rule calculated cells, "Margin %", I got 1.1 million bytes in a cached Stargate View.
When I set AllRuleCalcStargateOptimization=F, the default, and restarted the server, the Stargate view was over 40 million bytes for the same view. "Margin %" is a C: level element with "Sales" and "Costs" for children. My View had "Margin %" alone nested row-wise, and when I expanded it, "Sales" and "Costs" were all zeroes in the first case, but fully populated in the 40+ million-byte case. The Stargate log I set up indicated that "Sales" and "Costs" should have been cached, but it was as if the Stargate view's definition included the non-rule calculated descendants not shown in the view, but the cache itself only had "Margin %" because AllRuleCalcStargateOptimization=T.
This might be the "unexpected null suppression dropouts". Refreshing the view several times did not update the zeros in the first case, but invalidating the cache, and refreshing an altered view with "Margin %", "Sales", and "Costs" in the rows retrieved those values….And created a 40+ million byte cached view in the first case.
------------------------------
Walter Coffen
Technology Manager
QueBIT Consulting, LLC
------------------------------
Original Message:
Sent: Mon May 08, 2023 08:53 PM
From: John O'Leary
Subject: 2.0.9.16 AllRuleCalcStargateOptimization cannot =T when ViewConsolidationOptimization = T?
I missed this earlier in the release notes for 2.0.9.16. Perhaps related.
------------------------------
John O'Leary
Original Message:
Sent: Mon May 08, 2023 09:04 AM
From: John O'Leary
Subject: 2.0.9.16 AllRuleCalcStargateOptimization cannot =T when ViewConsolidationOptimization = T?
This is easily reproduced and I expect it is of use to others.
I don't see anything in release notes about these settings but we are experiencing fundamental issues with them both set to T (As they have been).
Simple reproducable case:
Cube: 3 dimensions
1) Version: Actual, Budget, 'Var Actual to Budget' (Aggregation of Actual +1, Budget -1)
2) Acc: Account 1, Account 2
3) Measure: Value
Rule to override the version aggregation
#########
skipcheck;
['Var Actual to Budget'] = C: if(!acc@='Account 1',['Budget'] - ['Actual'],['Budget']*2);
feeders;
##
Results
The following screen shot views are fundamentally the same cube view (I 'saved as' to show 2 views together),
a) one with the expanded view - where the aggregate version is incorrectly showing the aggregate as defined in the version dimension. Trace rule on aggregate shows the correct calcs
b) one with the aggregate only - showing correct calcs
PAfE DBRW referencing 'Var Actual to Budget' via TM1RPTVIEW returns the wrong values. Edit the cell and the correct value is returned.
Setting AllRuleCalcStargateOptimization=F with ViewConsolidationOptimization=T returns the above to correct operation.
Regards
------------------------------
John O'Leary
------------------------------