DLP worries me in principle since each mirror is electromechanically moved in and out of the optical axis millions of times in a day (in about four hours of operation). My experience with a similar system (MEMS micro accelerometers) indicates that over time, dislocations and stresses build up, and possibly after about five years of operation (at just 4 hours a day), 1.5 billions cycles will have been sustained, at that level, one should expect some of the mirror to first lose their movement reversibility (with dislocations build up, hysteresis should ensue), and some may even fail, starting to show parts of the display dead.
I think that both systems, however use a light engine with a rotating mirror, though it is possible (if the number of microelements is multiplied by four) to avoid the rotating color whel in both systems. Mind you, the difficulty or increasing by a factor of four the number of pixels (and reducing their size) is breater with DLP than with LCos. Eliminating the rotating mirror also would require over frequency of operation to be quadrupled, making the driving electronics more costly, and more so for DLP than for LCos, IMHO.