«

I just updated my perl version to 5.8 from 5.6.1 and with no code
change in the area of my list creation.... but I get a different order
now...

Here is the code I believe to be in question:


my @details = ((basename($fi)),$na,$nu,$ve,$ls,$st);
push @{$LIB_ITEMS{$pr}{$li}{$asm++}},@details;

.... later in my code...

foreach my $asm (keys %{$LIB_ITEMS{$prj}{$lib}}) {
....
}


I know that you cannot rely on a specific order from keys stored in a
hash -- however -- I did get a consistant order one way in 5.6.1 now I
get a consistant yet different order now in 5.8

It appeared that it was alpabetical (for $asm) in 5.6.1. and now
reverse alphabetical in 5.8

? anyone else notice this ?

I can just use a sort command with my keys -- just thought it was odd
that it was different.

John

wrote:

> I know that you cannot rely on a specific order from keys stored in a
> hash -- however -- I did get a consistant order one way in 5.6.1 now I
> get a consistant yet different order now in 5.8

Correct.

> It appeared that it was alpabetical (for $asm) in 5.6.1. and now
> reverse alphabetical in 5.8
>
> ? anyone else notice this ?

Yes.

> I can just use a sort command with my keys -- just thought it was odd
> that it was different.

No it's not really odd, since it is documented. From 'perldoc -f keys':

The actual random order is subject to
change in future versions of perl, but it is
guaranteed to be the same order as either the
"values" or "each" function produces ...

It is probably the side-effect of some internal optimization.

--Ala

Ala Qumsieh wrote:
> wrote:
>
>> I know that you cannot rely on a specific order from keys stored in a
>> hash -- however -- I did get a consistant order one way in 5.6.1 now I
>> get a consistant yet different order now in 5.8
>
> Correct.

Yes, as of 5.8.0 there's a new algorithm.

See perl58delta.

http://search.cpan.org/~nwclark/perl...e_Enhancements

As of 5.8.1 the order is no longer consitant (to resist a particular
class of DoS attack).

See perl581delta.

http://search.cpan.org/~nwclark/perl..._Randomisation