"I want to keep the keys/values in the same order for every array element in an AoH ref."

You just cant do that. You could instruct data dumper to PRINT them in sorted order (local $Data::Dumper::Sortkeys=1;), or you could program your own print loop to print them in a certain order,

use strict; use warnings; 
my $data=

[
    {
        'lastname'  => 'Smith',
        'city'      => 'Chicago',
        'state'     => 'IL',
        'firstname' => 'Jim'
    },
    {
        'city'      => 'Cleveland',
        'state'     => 'OH',
        'firstname' => 'Susan',
        'lastname'  => 'Jones'
    },
    {
        'state'     => 'FL',
        'lastname'  => 'Waters',
        'firstname' => 'Sam',
        'city'      => 'Miami'
    }
];

my @order=qw/firstname lastname city state/; 
my $rowct=0; 
for my $row (@$data) { 
  printf "%5i \n",$rowct++; 
  for my $ordername (@order) { 
    printf " %20s %-20s \n",$ordername,$row->{$ordername}; 
    }
  }
[download]

As an interesting alternative to the qw/.../ you may find some interesting info about getting the field names from the sql request at Re: Field names from DBI? . that would allow you to specify the order via the SELECT statement, but it a two step setup described in setting up and executing the SQL. YMMV

So Data::Dumper is throwing me off and Tie::IxHash just might be keeping them in order as advertised. I'll write a loop to check that. Thanks for the lesson in D::D

—Brad
"The important work of moving the world forward does not wait to be done by perfect men." George Eliot

So Data::Dumper is throwing me off and Tie::IxHash just might be keeping them in order as advertised.

It's unlikely that Data::Dumper has anything to do with it. As others have written, my guess would be that selectall_arrayref() is returning a reference to an entirely separate hash it generates and populates internally. After all, how would this function "know" that you want it to use the hash referred to by $data to populate with the data it is to return? I can't use Tie::IxHash to produce an example ATM, but consider:

c:\@Work\Perl\monks>perl -wMstrict -le
"package Foo::Bar { sub TIEHASH {} }
 ;;
 my $data = {};
 tie %$data, 'Foo::Bar';
 print $data;
 ;;
 $data = return_anon_hash_ref();
 print $data;
 ;;
 use Data::Dumper;
 print Dumper $data;
 ;;
 sub return_anon_hash_ref { return { 'sorry' => 'charley' }; }
"
HASH(0x3bd06c)
HASH(0x3bd15c)
$VAR1 = {
          'sorry' => 'charley'
        };
[download]

Note that the addresses of $data in the two print statements are different, as are the contents of the referents. Nothing you do with Data::Dumper is going to change this.

Bottom line: Write your own data display loop or routine and use whatever explicit key order you want.

---

Give a man a fish:  <%-{-{-{-<

Probably $data = $self->dbh->selectall_arrayref is replacing the hash reference with a reference to a different hash that isn't tied.

But hashes don't have ordering, why do you need ordering? Just for printing?

Yes, I'm building a CSV file, but thanks to the comments above, I'm thinking of 1) writing my own loop, or 2) using Excel::Template. Funny how we get myopic and miss the forest.

—Brad
"The important work of moving the world forward does not wait to be done by perfect men." George Eliot

Hi bradcathey,

I'm building a CSV file

If you use selectall_arrayref, that'll pull all the results from the database into memory. I imagine that fetching the results row by row with DBI's fetchrow_arrayref, the return value of which can be fed directly into Text::CSV's print method, would be more efficient.

Regards,
-- Hauke D