Why is "undef" after passing not readonly anymore?

LanX has asked for the wisdom of the Perl Monks concerning the following question:

  DB<174> use Scalar::Util qw(readonly)
  
  DB<175> sub tst { print readonly($_[0]) ? "r" : "rw" }
 
  DB<176> tst 0
r
  DB<177> tst undef
rw
   DB<179> print readonly(undef) ? "r" : "rw"
r
[download]

Bug or feature???

Cheers Rolf

UPDATE: to answer my own question, readonly only tells me if I can modify an alias w/o problems within a sub. The original undef in the outer scope isn't changed anyway.

  DB<181> sub tst { $_[0] = 42 }

  DB<182> tst 5
Modification of a read-only value attempted at (eval 119)[/usr/share/p
+erl/5.14/perl5db.pl:640] line 2.

  DB<183> tst undef
 => 42                   # really?
[download]

Comment on Why is "undef" after passing not readonly anymore? Select or Download Code

Replies are listed 'Best First'.
Re: Why is "undef" after passing not readonly anymore? by Tux (Canon) on Sep 05, 2012 at 18:25 UTC
See Juerd's talk "When undef isn't". When it doesn't show, disable CSS. Enjoy, Have FUN! H.Merijn	[reply]
Re^2: Why is "undef" after passing not readonly anymore? by LanX (Saint) on Sep 05, 2012 at 18:29 UTC
only sparse slides, no deeper explanation? Cheers Rolf	[reply]
Re^3: Why is "undef" after passing not readonly anymore? by Anonymous Monk on Sep 05, 2012 at 19:50 UTC
Sorry, "you had to be there" :)	[reply]
Re: Why is "undef" after passing not readonly anymore? by philiprbrenan (Monk) on Sep 06, 2012 at 00:34 UTC
It's to permit autovivification: `use feature ":5.14"; use warnings FATAL => qw(all); use strict; use Data::Dump qw(dump pp); sub aaa($) {my ($a) = @_; $a->{a} = 1; say STDERR dump($a); } aaa({}); aaa(undef); aaa(0);` [download] Produces: `{ a => 1 } { a => 1 } Can't use string ("0") as a HASH ref while "strict refs" in use at C:\ +Phil\Perl\PerlMonks\undef2.pm line 8.` [download]	[reply] [d/l] [select]
Re^2: Why is "undef" after passing not readonly anymore? by LanX (Saint) on Sep 07, 2012 at 08:37 UTC
Hi Good point, but I think it's even simpler! There are so many cases where undef is occasionally passed as argument, that inner variables of a sub like $_[0] must be writable. Cheers Rolf	[reply]