Avoiding TI abort when using ViewCreateByMDX

[PavoGa]

We are moving into using ViewCreateByMDX pretty heavily. One issue we have run into is testing whether we are going to return a data set or not. If it does not, the TI aborts instead of just skipping over the METADATA/DATA procedures like a traditional view does.

One method is to do a CELLGETN against an intersection that would have data and execute the ViewCreateByMDX only if the value is <> 0.

This is popping up when using RUNPROCESS and everyone freaking out when the server log shows the abort. Working on it, but wondered what other ways people are handling this?

[macsir]

Hi, I had similar problem when using SUBSETCREATEBYMDX and then switch to function SUBSETMDXSET, which allows process goes to epilog. Maybe you can replace yours with ViewMDXSet as well?

[PavoGa]

Hi, I had similar problem when using SUBSETCREATEBYMDX and then switch to function SUBSETMDXSET, which allows process goes to epilog. Maybe you can replace yours with ViewMDXSet as well?

Talking ViewCreateByMDX, not SubsetCreateByMDX.

[lotsaram]

Hi Ty,

I hear you. We have come up against the same issue. IMHO this is a defect and the behaviour should be the same as for a native view (no error, skip straight to epilog).

In testing we found that the error below

Error: Prolog procedure line (0): Unable to open data source "CUBENAME".

... can be avoided if rather than doing NON EMPTY in the MDX Expression we apply ViewExtractSkipBlanks after the view is created.

The workaround to do CellGetN on some consolidated intersection representing the view is the one we are also using though.

[Steve Rowe]

Just a note to say there is an APAR associated with this
https://www.ibm.com/support/pages/apar/PH16537