Update: stay away from my erroneous regexes.

I'd solve it using a regex:

sub elide_japhy {
  my ($s, $e) = @_;
  return $s == $e ?
    $s : 
    ("$s $e" =~ /^(\d*)\d* \1(\d*)$/ and "$s-$2");
}
[download]

The way it works is this: if the lower and upper bounds are equal, just return the number; otherwise (assuming the upper bound is actually greater than the lower bound), we try to match a prefix in the lower bound that is also found in the upper bound (which could just be zero characters), and return the lower bound, a dash, and the remainder of the upper bound.

As an example, for 325 through 349, the prefix is '3', and the remainder is '49'; for 324 through 329, the prefix is '32', and the remainder is '9'.

Finally, we could micro-size it. It's not entirely golfed, but I get rid of intermediate variables.

sub elide_japhy {
  $_[0] == $_[1] ?
    shift :
    ("@_" =~ /^(\d*)\d* \1(\d*)$/ and "$_[0]-$2");
}
[download]

Oh, and the logic could be embedded in the regex itself:

sub elide_japhy {
  "@_" =~ /^(\d*)(\d*) \1(?!\2)(\d*)$/ ? "$_[0]-$3" : shift;
}
[download]

_____________________________________________________
Jeff[japhy]Pinyan: Perl, regex, and perl hacker, who'd like a job (NYC-area)
s++=END;++y(;-P)}y js++=;shajsj<++y(p-q)}?print:??;