I'm not sure what you mean by "adjusted to your file", as
it seems to work quite well on any given data. It isn't
perfectly random, as things aren't shuffled quite as much
as the splice approach, but they are shuffled to the point
of being random-looking with no obvious patterns. Since
we're not picking winning lottery numbers, I thought that
would be acceptable.
I can only presume that since the output of the comparator used by
sort does return different values for the same $a-$b pair,
this could really bust a gasket on some of the older
Perls which perhaps assume that this would not occur. IMHO,
this is a bug in Perl more than it is a bug in the program,
as no program is supposed to be able to dump core, unless,
perhaps, it uses the dump command. | [reply] |