Benchmark: timing 1000 iterations of dimmesdale PU, dimmesdale loop, perlish...
dimmesdale PU: 36 wallclock secs (34.87 usr +  0.09 sys = 34.96 CPU)
dimmesdale loop: 43 wallclock secs (40.63 usr +  0.13 sys = 40.76 CPU)
   perlish: 35 wallclock secs (33.47 usr +  0.10 sys = 33.57 CPU)