in reply to Re: Dumping opaque objects
in thread Dumping opaque objects

Sadly that doesn't appear to do what I want (though I'd love it to): the docs for $Data::Dumper::Freezer talk only about modifying the object passed in, not about a return value. And indeed, trying it (with a different name to avoid the existing Math::GMP::stringify method) shows it is called but has no effect on the Data::Dumper output:

% perl -MData::Dumper -MMath::GMP -wle '
  $Data::Dumper::Freezer = "zify";
  local *UNIVERSAL::zify = sub { warn "zify $_[0]"; $_[0]->isa("Math::
+GMP") ? "z$_[0]" : $_[0] };
  my $x = [ Math::GMP->new(1) ];
  print "x[0]=$x->[0]";
  print Dumper($x);
  print "x[0]=$x->[0]";
'
Name "UNIVERSAL::zify" used only once: possible typo at -e line 3.
x[0]=1
zify 1 at -e line 3.
$VAR1 = [
          bless( do{\(my $o = 40040032)}, 'Math::GMP' )
        ];

x[0]=1

[download]

If I change it to assign back to $_[0] instead, Dumper outputs garbage ("\undef") and the data structure is corrupted to have the new string instead of the original object:

% perl -MData::Dumper -MMath::GMP -wle '
  $Data::Dumper::Freezer = "zify";
  local *UNIVERSAL::zify = sub { warn "zify $_[0]"; $_[0]->isa("Math::
+GMP") ? "z$_[0]" : $_[0] };
  my $x = [ Math::GMP->new(1) ];
  print "x[0]=$x->[0]";
  print Dumper($x);
  print "x[0]=$x->[0]";
'
Name "UNIVERSAL::zify" used only once: possible typo at -e line 1.
x[0]=1
zify 1 at -e line 1.
$VAR1 = [
          \undef
        ];

x[0]=z1

[download]

Replies are listed 'Best First'.
Re^3: Dumping opaque objects
by syphilis (Archbishop) on Jan 24, 2022 at 10:58 UTC
    Hi hv,

    I was able to get a Dumper() that appears more meaningful (to me, at least) with: 
    ## dumper.pl ##

use strict;
use warnings;

use Data::Dumper;
use Math::GMP;

my $x = [ Math::GMP->new(101), Math::GMP->new(203), Math::GMP->new(116
+7) ];

print Dumper("@$x");

__END__

C:\_32\pscrpt>perl dumper.pl
$VAR1 = '101 203 1167';

    [download]
    However, I'm not familiar with Data::Dumper and I've no idea if that's of any use to you.

    Cheers,
    Rob
[reply]
[d/l]

      Thanks Rob, that's essentially the sort of thing I currently need to do, but it requires knowing the exact structure to be dumped and walking it myself. I also want to distinguish between GMP objects and simple integers - particularly to spot when something that should be GMP is not or v.v. - so I'd end up needing to write most of Data::Dumper myself. I had hoped somebody would already have done that for me (and that someone here would have used it).

      My CPAN-fu is minimal, but of the 2,520 matches on metacpan for 'dump', a quick scan of the first 100 threw up some maybes such as Data::Printer (which cares mostly about coloured output) and Data::Dump::Streamer (which cares mostly about structure recreatability via eval). I also spotted a cross-reference to Data::Rmap, which might offer a sane way to roll my own. I'll try some of those and see what I get.

[reply]
        I also want to distinguish between GMP objects and simple integers

        Yes - inside my own gmp-based modules I often want to know about such things.
        I normally use XS or Inline::C subroutines for this.

        In pure perl code, as you may already be well aware, the ref() function will go a long way to providing useful information about references and module objects. 
        use warnings;
use Math::BigInt;
use Math::GMP;

my $v = 9;
my %h = (foo => $v);

for (\%h, [$v], \$v, Math::GMP->new($v), Math::BigInt->new($v), $v) {
  ref($_) ? print ref($_), "\n"
          : print "Not a reference\n";
}

__END__

Outputs:

HASH
ARRAY
SCALAR
Math::GMP
Math::BigInt
not a reference

        [download]
        You can also use the B module instead of XS/Inline::C to determine whether a scalar is an IV or an NV or a PV or a combination thereof.

        If you want more info on any of these aspects, just ask.
        (I don't want to annoy you by banging on with stuff that you already know (or stuff that you're not particularly interested in).

        Cheers,
        Rob
[reply]
[d/l]

In Section Seekers of Perl Wisdom