I was recently interested about a rigorous demonstration of the exact equivalence between angular spectrum propagation and Rayleigh-Sommerfeld propagation integrals, as mentioned in Goodman's "Introduction to Fourier Optics". I followed Goodman's reference for that: Sherman's "Application of the Convolution Theorem to Rayleigh's Integral Formulas" (1967)
In that publication, the second RS solution has a negative sign (equation 3), whereas in Goodman there is a positive sign. The positive sign seems logical from Goodman's demonstration with alternative Green's functions, but the negative sign leads to a perfect equivalence of the two RS solutions in the paper... Is there a mistake, or a missing step somewhere?