As the community investigates more complex flows with stronger streamwise variations and uses more physically inclusive stability techniques, such as BiGlobal theory, there is a perceived need for more accuracy in the base flows. To this end, the implication is that using these more advanced techniques, we are now including previously neglected terms of O(Re²). Two corresponding questions follow: (1) how much accuracy can one reasonably achieve from a given set of basic-state equations and (2) how much accuracy does one need to converge more advanced stability techniques? The purpose of this paper is to generate base flow solutions to successively higher levels of accuracy and assess how inaccuracies ultimately affect the stability results. Basic states are obtained from solving the self-similar boundary-layer equations, and stability analyses with LST, which both share O(1/Re) accuracy. This is the first step toward tackling the same problem for more complex basic states and more advanced stability theories. Detailed convergence analyses are performed, allowing to conclude on how numerical inaccuracies from the basic state ultimately propagate into the stability results for different numerical schemes and instability mechanisms at different Mach numbers.