Re^4: References for ano subs fixed at compile time? (nope)

You missed the point, the different calls to sub {1} print different refs, even that 1 is always the same code and not a closed over var.

The reason is that sub {2} reused the first ref and $b inhibits its release!

  DB<161> ;{ my $b; for (1..3){ print sub {1} ; print $b= sub {2} if $
+_==1 } }
CODE(0x86c5af0)
CODE(0x86c5af0)
CODE(0x86c5be0)
CODE(0x86c5be0)

  DB<163> ;{ my $b; for (1..3){ print 0+sub {1} ; print 0+($b= sub {2}
+) if $_==1 } }
141319216
141319216
141319552
141319552
[download]

Cheers Rolf

( addicted to the Perl Programming Language)

Comment on Re^4: References for ano subs fixed at compile time? (nope) Select or Download Code

Replies are listed 'Best First'.
Re^5: References for ano subs fixed at compile time? (opnode) by tye (Sage) on Jun 18, 2013 at 22:59 UTC
Well, that part is the stupider part. If address X is no longer used, it is obvious that address X might be re-used. If address X gets re-used, it is obvious that it might still be being used and so can't be re-used a third time. But, indeed, even completely identical code refs get recreated each time in my version of Perl (so I lose my bet). Which probably means that a closure is generated even if nothing got closed over. But each closure contains a pointer to the code that got compiled for that (trivial) sub. Perhaps that is what you want? - tye	[reply]
Re^6: References for ano subs fixed at compile time? (opnode) by LanX (Saint) on Jun 19, 2013 at 00:30 UTC
> Perhaps that is what you want? I thought I could use the code-ref as reliable identifier for the call position ( not just the line-number caller provides ). Something which can be used to emulate position dependent built-ins like the flip-flop operator. I tried `B::svref_2object` but the docs to B are somehow cryptic... Cheers Rolf ( addicted to the Perl Programming Language)	[reply] [d/l]