Comment on gauge choices and physical variables in QED

We consider possible definitions of physical variables in QED. We demonstrate that the condition $\partial_i A_i$$=0$ is the most convenient one because it leads to path integral over physical components with local action. However, other choices, as $A_3=0$, are also possible. The standard expression for configuration space path integral in $A_3=0$ gauge is obtained starting with reduced phase space formulation. Contrary to the claims of the paper [M.Lavelle and D.McMullan,Phys. Lett. B316 (1993)172] the $A_3=0$ gauge is not overconstrained.