Very neat program bikeNomad. I spotted some minor bugs/optimisations ... I must remark that that was no thorough testing, so I have probably left some more things to discover for other ;-)

delete($opts{l}); ## was: delete($opts{L});
[download]

typo ... otherwise there is an infinite loop with the l-option

# was: if (/@(?:([[:word:]:]+)::)?ISA\s*=\s*(.*)\s*/) 
if (/(?<!\\)@(?:([[:word:]:]+)::)?ISA\s*=\s*(.*)/)
[download]

I added a negative look-behind at the beginning to not allow a backslashed '@' in front of the ISA - so now even if run on the sourcecode itself (not that this is very sensible :) it produces no errors. Furthermore I removed the '\s*' at the end, this was unnecessary as everything is ate up anyway by the (greedy) '.*' .

while (!/;/)    # accumulate ISA value for multiple lines
{
    $_ = <$f>;
    $tmp .= $_;
}
@isa = eval $tmp;
# was: if ($@) { warn "Unparseable \@ISA line: $_"; next };
       if ($@) { warn "Unparseable \@ISA line: $tmp"; next };
[download]

The reason for creating the eval-error can be in any of the lines read in, not only in the last. Apart from that the error handling prints now always only the part after the equal sign.

-- Hofmator