Iowa doesn't smell right

[mr_lebowski]

88. Reports can definitely be wrong w/o the underlying data being questionable ...

Reports are 'windows' into underlying data, and the parameters of that window can be coded incorrectly.

In fact it's very common to discover reporting errors by comparing them with the raw data (which one assumes to be correct) and finding that things don't jibe between them.

Report code that fails to account for unexpected dupes in the underlying data is a REAL common mistake, for example. You do your tests on the report w/no dupes, but don't realize that the underlying table structure holding the data doesn't preclude dupes w/appropriate unique indexes.

Real people (not testers) go to use it and they create dupes that the testers didn't think to do themselves. Then the reports blow up.

Granted in this particular example there's an underlying data integrity issue in a manner of speaking, but it's very reparable. And this is just a for instance, in many cases, there wouldn't be one at all, only bad report coding.