On Tue, Nov 05, 2002 at 06:53:38PM -0000, Soren A wrote:
> Hello,
>
> In ExtUtils::MM_Unix, in the sub pm_to_blib(), there is:
> if ($l + $la + $lb + @a / 2 > 200) { # limit line length
> _pm_to_blib_flush($self, $autodir, \$r, \@a, \$l);
> }
> push @a, $pm, $blib;
> $l += $la + $lb;
> }
> _pm_to_blib_flush($self, $autodir, \$r, \@a, \$l);
> Why is the list of parameters being passed to _pm_to_blib_flush()
> with "$self" as the first argument? That makes "$self" appear TWICE
> to the latter sub. Once, implicitly, as in any method invocation;
> and a second time because of the explicit argument.
I don't think that your analysis is correct.
_pm_to_blib_flush is being called as a regular subroutine. In perl5 there
is no implicit invocant in $_[0] if a sub is called as a regular subroutine,
even if it functions as a method in a class.
I believe that the code above is written that way to fake a method call of the
form
$self->_pm_to_blib_flush($autodir, \$r, \@a, \$l);
which does end up with the invocant object as $_[0] inside the subroutine.
> Is this not erronious code? It caused an error caught under the
> "strict 'refs'" conditions when I ran it; because a later arg
> that is supposed to be a string scalar is used instead as a
> scalar ref. I mean, the args all out of joint.
I'm not sure if it is erroneous code; I don't know if MM_Unix (or any part
of MakeMaker) is expected to work under use strict; It might be that a
soft reference was intended.
The person who probably knows best is Schwern, and he's currently out of
contact. This is a big reason this list is rather quiet currently.
I hope that either I removed the spam protection in your e-mail address
correctly, or that you are subscribed to the list, else you may never
see this reply.
Nicholas Clark
--
INTERCAL better than perl? http://www.perl.org/advocacy/spoofathon/
