Is it true that if I have two CW-spectra E and F that I can define E \smash F as the spectra which in degree n is hocolim_{i,j} \Sigma^{n-i-j} (E_i \smash F_j) , where i and j ranges over the integers and I take the hocolim in the Quillen model structure of spaces. Of course, I can define it that way, but does it agree with the old-fashioned way of defining the smash product of spectra ?