I did some calculation about depth of field, combine the result provided by Rubén Osuna and Efraín García (see here
http://www.luminous-landscape.com/tutorials/resolution.shtml )I found an interesting conclusion:

Assume that the sensors all have infinite resolution, and the lenses are all diffration limited, then In order to have the same image(composition, DoF, FoV), we would result in the
same image resolution
.

Take 35*24mm sensor and 48*36mm sensor for example:
35*24mm 48*36mm
lens for same fov: 50mm 75mm
f-stop for same dof: f11 f16
diffraction limited resolution of image: 15mp 15mp

This result is particularly useful if we shoot macro, which we have to pull down the f-stop to obtain enough dof.

Interestingly, if you use a 20mp 35mm Dslr and rarely shoot f8 below, then the big sensor altanative(dont know how to spell,sorry..) doent seem to have much advantage.

Anyway all these are just based on maths and is highly theoretical conclusion. I would like to do some real world test when I could, I also wish that someone is interested and do some test too(I still don't have a DB myself ).

regards,
Dustblue

You are right.

Now, consider the dificulties of making a diffraction limited lens. Aberrations depend on FoV, aperture and format size.

I agree.
But consider when shooting macro withoud lens/back movements, all the lenses are diffraction limited, and thus the sensor size become less important.

Is there anyone who shoot lots of jewry/watch/insects have some experience about this? I think if this is proved to be true, the delicated macro photographers could save a lot money.