sub get_parents {
  my (@p, $o); 
  for (grep !/^$/, split ("/",$_[0])) {push(@p,"$o/$_");$o.="/$_"}
  return \@p;
}
[download]

print join(" ", map { sprintf "%#02x", $_ }unpack("C*",pack("L",0x1234
+5678)))
[download]

I suggest avoiding /^$/ because regexes are a lot more expensive than calling length. Thus:

for (grep { length } split(m{/+}, $_[0])) # ...

    -- Chip Salzenberg, Free-Floating Agent of Chaos

Another reason to avoid /^$/ is that it non-intuitively matches a single newline:

#!/usr/bin/perl -wT
use strict;

# only one one of these is an "empty" string right?
my @arr = ("", "\n", "dog", "cat");  

my @haslength = grep {length} @arr;
my @notempty  = grep {!/^$/} @arr;

print scalar @haslength, " members have length\n";
print scalar @notempty,  " members are not \"empty\"\n";

__END__

=head1 OUTPUT
3 members have length
2 members are not "empty"
[download]

So, /^$/ actually classifies a single newline as an "empty" string.... Sometimes thats what you want, but it isn't usually what people expect.

-Blake

given that the grep for /^$/ was to skip the null first field returned by split (because the first character in the string was the delimiter "/"), wouldn't it be "better" (more readable) to just explicitly remove it- ie

s#^/##; # remove the leading delimiter
[download]

The code as written silently munches all null fields which is fine because we don't expect a directory called null (ie in /foo/bar//rab/oof) but if the idiom is transported to another set of data...

split /' '/ nicely removes any leading white space... it could be handy to have an option to do this for other delimiters.

--
my $chainsaw = 'Perl';