When you do that today, you get an error: "Can't coerce %s into hash".
As far as Perl 5 is concerned, this is consistent with not vivifying
an undefined variable, since that doesn't have a hashref type
either.[1] So Peter's request means introducing an inconsistency, and
may require some sort of option to control which way the inconsistency
leans toward.
[1] In Perl 6, there are things called prototype objects, which are
typed undefined values.
On Jan 21, 2008 2:16 PM, Jason Elbaum <[EMAIL PROTECTED]> wrote:
> Why should it be any different from $q->{xyz} if $q already has a
> reference of another type?
>
>
>
> On Jan 21, 2008 2:12 PM, Gaal Yahas <[EMAIL PROTECTED]> wrote:
> > Then you'd have to have another option for making that option do
> > different things if $q already had a reference of another type...
> >
> >
> > On Jan 21, 2008 11:30 AM, Peter Gordon <[EMAIL PROTECTED]> wrote:
> > >
> > > So shouldn't there be an option to allow %$q and @$q to auto-vivify?
> > >
> > >
> > >
> > > > On Mon, 2008-01-21 at 01:14 -0800, Yitzchak Scott-Thoennes wrote:
> > > > On Mon, Jan 21, 2008 at 10:59:52AM +0200, Peter Gordon wrote:
> > > > > This is a bit off topic - but anyway ....
> > > >
> > > > Seems on topic to me?
> > > >
> > > > > Can someone explain, what appears to me to be bizarre behaviour.
> > > > >
> > > > > Normally, Perl won't produce runtime fatal errors if the syntax is
> > > > > correct.
> > > >
> > > > Yes, it does; all the time. Particularly with references.
> > > >
> > > > > Why can't the two lines marked below just return undef.
> > > >
> > > > I'd expect an empty list, if anything. undef would be very wrong.
> > > >
> > > > > The failure on @$q is particularly irritating. If a reference to an
> > > > > array is passed and the value happens to be undef because there were
> > > > > no
> > > > > entries, you need to either check that the value is a reference to an
> > > > > array, or to do an eval. Good programming practice would say the the
> > > > > original value should be initialized correctly. But I would prefer
> > > > > Perl
> > > > > to just me undef.
> > > > >
> > > > > And if you are going to argue that this is correct behaviour, then why
> > > > > does $s->{aaa} and co work?
> > > > >
> > > > > use strict ;
> > > > >
> > > > > my $s ;
> > > > > my @t = %$s ; # <-- This causes a fatal error
> > > > > my $t1 = $s->{aaa} ; # This works
> > > > >
> > > > > my $q ;
> > > > > my @w = @$q ; <-- This causes a fatal error
> > > > > my $t2 = $q->[10] ; # This works
> > > >
> > >
> > > > Basically, some ways of using a reference will promote ("vivify")
> > > > undef to a reference of the correct type, and some will not.
> > > > Mostly, the division is in whether you are potentially modifying
> > > > the referred-to structure or not. Specifying a hash key or array
> > > > index potentially creates it, so counts as modifying (even though
> > > > in the simple case you show, no modification would take place,
> > > > element dereferencing in general *may* modify).
> > > >
> > > > You can just initialize your variable to start with:
> > > >
> > > > my $s = {};
> > > > my $q = [];
> > > >
> > > > or check if it is undef and, if so, provide a suitable empty hash or
> > > > array:
> > > >
> > > > my @t = %{$s || {}};
> > > > my @w = @{$q || []};
> > > >
> > > > or even initialize it on the fly:
> > > >
> > > > my @t = %{$s ||= {}};
> > > > my @w = @{$q
> > > >
> > > >
> > > > _______________________________________________
> > > > Perl mailing list
> > > > [email protected]
> > > > http://perl.org.il/mailman/listinfo/perl
> > >
> > > _______________________________________________
> > > Perl mailing list
> > > [email protected]
> > > http://perl.org.il/mailman/listinfo/perl
> > >
> >
> >
> >
> > --
> > Gaal Yahas <[EMAIL PROTECTED]>
> > http://gaal.livejournal.com/
> >
> > _______________________________________________
> > Perl mailing list
> > [email protected]
> > http://perl.org.il/mailman/listinfo/perl
> >
> _______________________________________________
> Perl mailing list
> [email protected]
> http://perl.org.il/mailman/listinfo/perl
>
--
Gaal Yahas <[EMAIL PROTECTED]>
http://gaal.livejournal.com/
_______________________________________________
Perl mailing list
[email protected]
http://perl.org.il/mailman/listinfo/perl
