Data::Dumper: Show only existing hash keys returned by Sortkeys(sub{...})

[jimav]

This is an enhancement suggestion.

Currently a hash member is shown for every key returned by a Sortkeys sub even if there is no such member. The output shows non-existent members as existing with an undef value.

It would be nice (and arguably more correct) to omit hash members which don't actually exist.

To get a correct display, one currently must do something like

Data::Dumper->new([$href])->Sortkeys(sub{ [ grep{ exists($_[0]->{$_}) } qw/all possible keys/]}->Dump

Whereas it would be nice to be able to just say

Data::Dumper->new([$href])->Sortkeys(sub{[qw/all possible keys/]}->Dump

Demo program:

#!/usr/bin/perl
use strict; use warnings; use feature qw/say/;
use Data::Dumper;

my $href1 = { aaa => 100 };
my $href2 = { aaa => 100, OPTIONAL => undef };
say Data::Dumper->new([$href1, $href2],['href1','href2'])->Sortkeys(sub{[qw/OPTIONAL aaa bbb/]})->Dump;

Current results:

$href1 = {
           'OPTIONAL' => undef,
           'aaa' => 100,
           'bbb' => undef
         };
$href2 = {
           'OPTIONAL' => undef,
           'aaa' => 100,
           'bbb' => undef
         };

Desired results:

$href1 = {
           'aaa' => 100,
           'bbb' => undef
         };
$href2 = {
           'OPTIONAL' => undef,
           'aaa' => 100,
           'bbb' => undef
         };

[jimav]

I'm sorry to not make clear this is about Data::Dumper.

The 'Issues' link for Data::Dumper at https://metacpan.org/pod/Data::Dumper pointed here so I didn't realized until later that this was for all Perl core issues.

[demerphq]

FWIW, I don't think we can do what you ask for without introducing a new callback for this. If we added a SortExistingKeys callback then we could add this behavior, but changing the Sortkeys() callback like this IMO would be a regression, I have used this to "fill out" missing keys in a hash before, and I think others would be too. So where you see an annoyance, others might be expecting this. I think both behaviors are reasonable, the problem is changing in mid flight like this.

Maybe there is another way to solve this, perhaps a second argument to Sortkeys() could work?

[jimav]

Hmm, I see. Probably not worth the pod bloat to add another parameter or new callback.