Agreed to your post. IF the wall roughness is small relative to the wavelength then it will not be much of an effect.

I was thinking more about how one could use a deliberate roughness at the wall to shape the wavefront by creating a boundary layer effect. from my book it is obviuos that one wants the wavefront to diminish at the edges of the waveguide.

Also, please remember that I do not use waveguides or horns at LF. They need to be too big to do any real good. Sure they offer some gain, but remember that I always use big high performance drivers for directional reasons and so gain is not really very important to me. (True unless I am doing a pro sound system, but then the waveguides can be big.) I use waveguides strictly for their directivity qualities, not for their loading ability. SO when I think about horns or waveguides I never think about low frequencies. In essence, if the wavelength is bigger than the waveguide dimensions then the waveguide is no longer a directional control device and I don't really consider it.