Dave's Adventures in Business Intelligence

Hi folks… a few days ago I posted an update via a comment on my universe compare tool page regarding a small bug. It’s not a huge deal, otherwise I am sure someone would have reported it by now. If for some reason you had a context that had zero joins then the program would not properly detect that and move on without generating an error. I posted what turns out to be a partial fix in the comment area.

While validating that change, I discovered that there are in fact two places where the code has to be updated in order to properly detect an empty context. I also found out that the process was not really finding changes in class properties at all! Oops. 😳 In order to handle the recursive nature of classes (class can contain sub-classes which can contain sub-classes and so on) the output row number for the target worksheet is passed as an argument. It seems that the first row was then cleared out, leaving the actual change log data to start on row 3 rather than row 2. (Row 1 is the column headers.)

Later on when the “detect changes” algorithm is called, it looks for a non-empty value in row 2. Given that this was never true for the class extract, changes in classes were never properly documented. The reason I suspect nobody called this out as a bug before is because the object change detection process was already capturing changes in class names, so at least that part was there. But if there were changes in class help text or visibility attributes or other class properties they were not being captured.

Given that there are several changes coming, I will be posting an update to the downloadable code within the next week or so.

As of today I don’t have any time to even make plans to create a similar version of this utility for .UNX universes created via the Information Design Tool.