Front page | perl.perl5.porters |
Postings from May 2010
On Mon, May 03, 2010 at 05:27:56PM +0100, Dave Mitchell wrote:
> On Thu, Apr 29, 2010 at 10:51:26AM -0700, Eric Brine via RT wrote:
> > A simpler demonstration:
> >
> > ---- BEGIN CODE ----
> > #!/usr/bin/env perl
> >
> > use strict;
> > use warnings;
> >
> > {
> > package Tie;
> > sub TIESCALAR { return bless({}, shift); }
> > sub STORE { next; }
> > }
> >
> > tie my $x, Tie::;
> > { tied($x)->STORE(123); } # warns, but works
> > { $x = 123; } # warns and dies
> > ---- END CODE ----
>
> I think this behaviour is correct. Tied method calls shouldn't consider
> any context outside the method call itself. After all, the tied action
> could have been invoked from almost anywhere, including somewhere deep
> within XS code.
>
> I also think the behaviour is deliberate. Magic method calls are run on a
> new set of stacks, and while die etc will pop back across multiple
> context stacks, pp_next etc only look at the current stack.
Yes, your explanation makes sense. I can't really see that this can be "fixed".
Nicholas Clark
Thread Previous
|
Thread Next