Yes. Each distance has to be a valid intra-block-pair distance, but they don't have to be the same as long as they're all [54..59]+31N -- we can design messy rephaser/rakes to run on pretty much any such spacing (I think -- haven't actually proved this for 90:90:90 yet, for example.)codeholic wrote:Is it necessary, that all trails should be separated by the same distance? Would 90:58:90 also work the same way, for instance?
To get a sample "messy RR" just take two trails off of any existing rephaser or rake. One side will still function just as it did before, but the side where you removed two trails will send out a backward glider stream. As long as something can catch these extra streams -- junk, forward gliders, whatever -- then we can do useful constructions with the output gliders on the other side. In particular, we could run a forerake and a backrake to set up another block trail N ticks away, on each side of the original four trails.
Now we have six trails. But the frame will have to switch: what used to be an intra-block-pair distance is now an inter-block-pair distance, and vice versa.
Next question: What if we start with just three trails, e.g., 90:90? We can _still_ run working rephaser/rakes on let's say the left side, but the right-hand lane will have extra blocks in it every time a Herschel climber goes by. That just means we have to get a single reversed lateral forward stream from somewhere lower down (or, less likely, somewhere higher up), to clean up those blocks before the next singleton Herschel climber comes along.
... No chance, I think. Those reflected forward streams are going to be more expensive to construct than the fourth block trail would be -- right?