Which is why I format the ST thus:

@a = map {
    $_->[0]
} sort {
    $a->[1] <=> $b->[1]
    ||
    $a->[ 2 ] cmp $b->[ 2 ]
} map {
    [ $_, COMPLEX EXPRESSION ]
} @b;
[download]

It makes the map / sort / map structure immediately apparent and allows for arbitrary complexity within the blocks, including multiple lines and even local variable declarations to be used without getting into that whole obfu-style thing.

A similar layout lends itself to the GRT:

@a = map{
    substr $_, 16
} sort map {
    m[...(\d+)...( \d+.\d+)...([a-z]4)...] or die "Illformed data '$_'
+";
    pack 'A4dNA*', $3, $2, $1, $_
} @b;
[download]

I never format single line if or else blocks one line either for same reason that doing so obscures the structure of the code.

I also like the consistancy with map/grep/sort chains and normal assignment. That is: result = manipulation of inputs

P6 has the pipe operators which allow you to reverse the flow src ==> tmp => dst, but I don't forsee me using that. I guess I grew up with assemblers, and even cpm/pip where dst = src was the way of things.