Schwartzian transformation is quicker only if the sort criteria is complex enough. ST makes sure the computation is only done N times instead of O(N*log(N)) times at the expense of creating a temporary array of two-item arrays containing the original values and the computed results. If the overhead of this AoA is bigger than the gain obtained by decreasing the number of "complex computations" ST is slower than the ordinary sort.
I think the -s is not a well chosen example for ST. Whether you actually gain or loose by using ST depends on too many things, the disk speed and cache size, the OS, the number of files in the directory, the speed of your CPU, the amount and speed of your memory, ...
IMHO of course, Jenda
Always code as if the guy who ends up maintaining your code
will be a violent psychopath who knows where you live.
-- Rick Osborne