In 1975 Doi and Edwards predicted that entangled polymer melts and solutions can have a constitutive instability, signified by a decreasing stress for shear rates greater than the inverse of the reptation time. Early experiments did not support this, and more sophisticated theories were developed that incorporated Marrucci's idea (1996) of removing constraints by advection; this produced a monotonically increasing stress and thus stable constitutive behavior. Recent experiments have suggested that entangled polymer solutions may possess a constitutive instability after all, and have led some workers to question the validity of existing constitutive models. Based on this intense interest we have revisited some of the phenemology present in state of the art tube models for entangled polymers, and performed calculations that take into account the stress inhomogeneity inherent in rotating rheometers (cone and plate and cylindrical Couette). Using the Rolie-Poly model with an added solvent viscosity, we show that (1) instability and shear banding is captured within this simple class of models; (2) shear banding phenomena is observable for weakly stable fluids in flow geometries that impose a sufficiently inhomogeneous total shear stress; (3) transient phenomena can possess inhomogeneities that resemble shear banding, even for weakly stable fluids. Many of these results are model-independent.