Re^3: do $n exist in grep?

[davido]

I use a similar idiom in formulating a return responses or external calls, sometimes:

my ($stdout, $stderr, $ret) = capture {
    system('foo',
        ($somearg  ? ('--somearg'  => $somearg ) : ()),
        ($otherarg ? ('--otherarg' => $otherarg) : ()),
    );
};

...or...

return {
    status => $s,
    ($optional_field ? (optional_field => $optional_field) : ()),
};

Caveat emptor. The need to do either of these may constitute code smell, but sometimes purity has to give way to get it done-ery.

[choroba]

It's also possible to shorten the lines with

('--somearg'  => $somearg ) x !! $somearg,
('--otherarg' => $otherarg) x !! $otherarg,

but only if the left hand side of x doesn't have any side-effects (because unlike the ternary operator, foo() x 0 still calls foo).

See also PerlX::Maybe.

[davido]

That's interesting. I'm not sure it qualifies as coding for clarity, but perhaps neither does my example.

[Marshall]

Interesting:

return {
    status => $s,
    ($optional_field ? (optional_field => $optional_field) : ()),
};

I personally would not make a hash key optional like that.
I would always return the hash key and then an undef value for that key if it is not valid.
But as is often said "there is more than one way to do it".
It could also be that I haven't thought this completely through and haven't considered what
the calling code looks like and how it processes this returned info.