If you are dealing with a commercial product and not a home made antenna then the spacing etc is already done and you are simply tuning for the best match. That is probably what 99.9% of the people do as very few people home brew a beam antenna. In any event you are not concerned with "phasing" of the parasitic elements as they are not driven. All you simply do is space them for maximum forward gain or best rear rejection whatever you want. That is simple with a field strength meter. After the gain/FB is peaked then you need to tune the feedpoint for best SWR. That is when you decide what type of matching system you will used based partly on what the unmatched feedpoint impedance is. Standing an antenna on it's reflector will not upset the FB or gain characteristics and allows a fairly accurate way to tune an antenna on the ground. A lot of beams are impossible to tune once in the air and MUST be tuned on the ground. Tribanders are a good example of that. It is impossible to reach the ends of the elements to adjust the lengths of tubing when the ends are 15 or 20 feet from the tower. Spending time mounting and unmounting and remounting the antenna a dozen times to do so is out of the question. On all my ham antennas I have gone with the manufacturers suggested dimensions and have always had thiongs work out. Who cares if the SWR is 1.5 or 1.6:1 at the band edges on 20m?
CK is absolutely right about the adjusting business up high, it is not practical for the reasons he noted even though I noted the idea of tuning at installed height was best.
I went back and checked again, and the practical difference when standing the beam up close to the ground vs. a 36' for example made very little difference in the match as, so he's right on that too. I used a model that came with the Eznec software and it was tuned for resonance only. But, it was not showing the proper phase, and that is what got me off into talking about phase.
CK, this model had no directors, it was a simple two element yagi with reflector and driver elements only. I didn't intend to leave the thought that this idea about phasing was anything but hypothetical.
When I adjusted the model for the best resonance, spacing, and element length I also saw improvement in the match and the reflector showed to be out of phase, which I think it should show, and the driven element then showed to be in phase, also like I would expect. You might agree with this much regarding phase?
I said in the beginning of my earlier remarks this was for consideration only, regarding the importance of phasing. But, in looking back the difference indicated very little difference after all was said and done. So you are correct again.
On second thought guys, I would not overly worry about this idea on phasing, even though a multi-element beam is a phased device.
My original idea was to see what Eznec might show when placing a beam on its reflector and the front end straight up in the air. I did that and the results supports what CK said about the idea of tuning with the beam sitting on or near the ground.
I jumped the gun guys, not to worry about phasing as long as your antenna is responding as you would expect with match, reflection, and whatever you can tell with forward gain...I wouldn't worry about phase as CK notes unless maybe the beam acts more like an omni, and then the lack of proper phasing might be the problem.