The important thing to know about Perl unicode support is that Perl does not track the type of a scalar. You, the programmer, need to keep track of whether you have a string of bytes or a string of unicode characters. The easiest way to do this is always decode bytes (like utf8 or utf16) into characters the moment it enters the program, like with your ":encoding(UTF-8)" mode.

As it happens, the decode_json function expects bytes as input, assuming you haven't done the decoding yet, and then it both decodes UTF-8 and parses JSON at the same time. On the other hand, if you say JSON->new->decode($string) that assumes you provided it with a unicode string.

So in summary:

open my $fh, '<', $filename;
$bytes= <$fh>;
$data= decode_json($bytes);
[download]

or

open my $fh, '<:encoding(UTF-8)', $filename;
$chars= <$fh>;
$data= JSON->new->decode($chars);
[download]