Tests for the “falseness” of the return-value are used in my code only when the value is 1 or 0.
I more or less agree with the rest of what you are saying, but I do not agree with that. It is perfectly legitimate, Perlish and idiomatic to evaluate an array in scalar/boolean context to figure out if it is empty. The scalar return value may be 0 (the array is empty) or 1, 2, 3, ..., any positive integer other than 0 (the array is not empty). For example:
while (@array) {
my $val = shift @array;
# do something with $val
}
or possibly:
while (my $val = shift @array) {
# ...
}
or:
if (length $string) {
# ...
}
(but not if ($string), because that might not give the expected result with the string "0"). Or you could have a sub return undef (false) if it did not succeed, or a hash or array reference (something that is always true), or even a simple array or hash, if it did succeed.
Similarly, if you open a file with this idiomatic syntax:
open my $fh, "<", $file or die "failed to open $file $!";
you're not testing for 0 or 1, and yet it works perfectly and is considered to be a best practice.
Of course, you've gotta know what you're doing and should probably try not to be too clever. I once got bitten by something like this:
$error_count++ and next if some_condition();
which did not work correctly the first time through the loop, because $error_count was 0 (or undef, whatever it was, I don't remember, it happened some years ago). While:
++$error_count and next if some_condition();
would have worked fine, I was probably trying at the time to be a little bit too clever (and too concise).
Limiting boolean tests to only 0 and 1 values is taking off a lot of Perl's expressive power.
Update: crossed out one example after the comment below from Anonymous Monk.
|