Thanks, Dalia. A couple of suggestions... In the code where you explain that we're now looking for both getaStart and getAStart, could you include the JavaBeans section as a reference? Down the road, if somebody looks at this code, they are going to wonder why this "exception" was put in place. Also, the text above made me think about another scenario... What happens if the attribute is named a_Start? That is, the second character is not an alphabetic character? Does the JavaBean spec outline this behavior? And, how would we respond? Other than that, this looks like a good patch. We're fixing this edge case, but leaving the current behavior.
On a related note... What happens when we generate Entity classes? If we read an attribute from the database named aStart, which getter/setter methods get generated? Is this a completely separate code path and, thus, your code changes don't affect that processing? Or, are they related and now we're (accidentally) changing how the Entity classes are getting generated?