Ooh, nice! That's eminently readable. Much more likely to survive a code review than aristotle's ex-bang-bang operator.

Rather than

my @part = (
    'http://example.net/app',
    ( 'admin'  ) x!! $is_admin_link,
    ( $subsite ) x!! defined $subsite,
    $mode,
    ( $id      ) x!! defined $id,
    ( $submode ) x!! defined $submode,
);
[download]

we would now have

my @part = (
    'http://example.net/app',
    grep( $is_admin_link.    'admin' ),
    grep( defined($subsite), $subsite ),
    $mode,
    grep( defined($id),      $id      ),
    grep( defined($submode), $submode ),
);
[download]

I like it.

What is the difference?

None.

I dare say the ex-bang-bang approach is faster, but then I would be rightly accused of premature optimisation. The grep approach, while slower (but at a cost approaching zero) is more easily figured out from first principles.

The effect of x!! is not easily discerned; one has to write a harness to explore its behaviour. Either because one has encountered a similar idiom in a different language and wants to see if Perl matches expectations based on prior experience, or else they have not the faintest idea of what it does (and once one sees what it does, can they determine why?)

So in summary, there is a difference. One is tricksy, the other is merely degenerate.

• another intruder with the mooring in the heart of the Perl