We are trying to upgrade the SAP Connector from v6.0 to v6.2 (could be v6.3 but we heard that it has some bugs...), but we noticed that some fields, that once were interpreted as numbers, are now being interpreted as texts.
We discovered that starting from v6.1, Qlik has modified the behavior of the query engine and the resulting dataset: "The SAP data types CHAR (character) and NUMC (numeric character) were previously interpreted (in the Qlik script) as ‘numeric’ if they contained only digits. Leading zeroes were removed. (0000141000 became 141000). Now they are always interpreted as ‘text’. Leading zeroes are kept. These data types are not often used in calculations, but if they are the new behavior might require script changes since calculations expects numeric values" (extracted from Release Notes v6.1).
I've attached a QVW comparing the resulting QVDs from VBAP table. The production environment is on v6.0 and test on v6.2.
We have a big 3-tier (extraction, transformation and viz apps) environment, and until now identified two options:
1. Isolate the solution in the extract layer, applying the number interpretation function in the sql queries so that the resulting QVDs remain the same as the old QVDs from v6.0: we consider this as a workaround, but easier to measure the effort. All the new queries would have to have this interpretation workaround to match with the rest of the data.
2. Assume that the query results are now "correct", and adjust all the transform and viz apps tiers to reflect this change: this option is "cleaner" (conceptually there would be no workaround involved), but more difficult to measure the effort. Probably we would have to do this by trial and error, adjusting the problems as they are identified.
Now we have to decide which way to go.
Can you share your thoughts? Any help appreciated.
Thanks in advance
VBAP.qvw 3.5 MB