Dirac erroneously tried to impose space-time symmetry on the time-skewed Schroedinger equation, which is the time component of a Lorentz-covariant four-vector system of equations -- that system's three space-component equations specify the quantum three-momentum operator in coordinate representation. Dirac's misconception resulted in a noninteracting-particle Hamiltonian that isn't the time component of a Lorentz-covariant four-momentum times c, and which causes the noninteracting particle to spontaneously undergo immense acceleration of the order of c squared divided by the particle's Compton wavelength, and to also have a fixed unphysical speed which is c times the square root of three. Dirac's Hamiltonian has a physically untenable unbounded-below set of negative energy eigenvalues, which have been airily "reinterpreted" as (very questionably) implying propagation backward in time. Dirac's misconceived Hamiltonian is in any case irrelevant since a noninteracting particle's Lorentz-covariant four-velocity times its mass m times c has a time component which is a superbly-behaved Hamiltonian with a simple space-time propagator for quantum wave functions. Via a Lorentz-invariant action integral, Lorentz long ago extended this noninteracting-particle Hamiltonian to describe the particle's interaction with an electromagnetic four-potential. Here we modify Lorentz's Lorentz-invariant action integral to accommodate the spin-1/2 particle by adding the Lorentz-invariant extrapolation of the nonrelativistic spin-1/2 particle's magnetic-moment potential energy in a magnetic field. We also point out the important fact that when particles can be produced, perturbation contributions become increasingly invalid with increasingly high virtual momentum values, which must be cut off.