TM1 Forum

We believe we have identified what seems to be a bug in PA2.0.

If HierarchyAttrPutS is used to update a dimension caption AND is followed by a subsequent SubsetCreateByMDX, the caption update goes away. The only way to prevent this is to execute a CubeSaveData(}DimensionAttributes) before the next SubsetCreateByMDX.

There may be a potential other workaround we have not tested, but for now we are seeing the behavior with HierarchyAttrPutS and SubsetCreateByMDX as repeatable. We have not tested with SubsetCreate.

Is the subset created with SubsetCreateByMDX a temporary subset or a permanent subset? Sounds like you're possibly running into an object locking issue (if the subset is permanent), which leads me to believe it is more of a technical limitation due to TI commits more so than a "bug". I agree it's frustrating, but without "hacking" your way to a commit (i.e. CubeSavaData: which has it's own negative sides; see below thread), I'm not sure there would be much of a way around it.

http://www.tm1forum.com/viewtopic.php?t=12417

Most are temporary. Have not tested if there is a difference.

Yes, have read that thread before and do not see any negative sides to using CubeSaveData in our situation. That thread actually recommends CubeSaveData over SaveDataAll when needing to commit cube changes.

But this is what is odd. I could see the object locking applying when the MDX is executed prior to the H...AttrPutS, but not after. We are seeing this when the MDX has nothing to do with the dim container or hierarchy for which the caption is updated.

I do not recall ever having this with AttrPutS. Does not seem to affect HierarchyAttrPutN, but I'd need to confirm that again. We've moved on with the workaround, but I still think it is a bug.

At some point, I'll try some thing to further refine the scope of the problem.

BTW, the other workaround I want to try is updating the caption attribute in }ElementAttributes_}Dimensions and using a rule to populate it in }DimensionsAttributes.

Hi

I suggest that you try the following in the order given to see if any of them works around this.

a) Swap the order of the attrputs and the SubsetCreateByMDX so the AttrPutS comes second
b) Do the SubsetCreateByMDX on the Prolog and the AttrPutS on the Epilog
c) Put the SubsetCreateByMDX and the AttrPutS in separate processes and call them from a top level process
d) Put them both in separate processes and trigger them separately eg by a Chore

Regards

Paul Simon