With the "_from_file" method, you need to be reading from a disk file.

With the loader method, you can load image data which is in a scalar string, or slurped in from a disk file, often referred to as blobs in ImageMagick terminolgy. ( See Convert Gnome2::Canvas::Pixbuf to Image::Magick Array ). This is often quite handy, like when you pull an image in from the web, and want to display it, without first writing it to a disk file. Like:

  my $graph_image = get($http_request_string);
 $graph_image at this point is a string, with a PNG-looking header.
 my $loader = Gtk2::Gdk::PixbufLoader->new;
  $loader->write($png_data);
  $loader->close;
  my $pixbuf = $loader->get_pixbuf;
[download]

Even though there is Gtk2::Gdk::Pixbuf->new_from_data, it does not automatically handle autodetection of file type as well as the Gtk2::Gdk::PixbufLoader. In other words, the Pixbufloader will autodetect from it's list of recognized types, making it the best method to use if you want an easy load.

Here is a simple way to list the autodetected file types which your Gtk2 build recognizes.

#!/usr/bin/perl
use strict;
use warnings;
use Gtk2;

my @formats = Gtk2::Gdk::Pixbuf->get_formats();
my @exts;

foreach my $format ( @formats ) {
    foreach my $key ( keys( %$format ) ) {
           next unless $key eq 'extensions';  
         
       foreach my $elem ( @{ $format->{ $key } } ) {
                push @exts, $elem;
            }
     } 
 }
print "@exts\n";
[download]

And finally, [the original message from [aristotle]:
<p><quote aristotle>

OT: I like to partition things like into smaller scopes:

    my $pixbuf = do {
        my $loader = Gtk2::Gdk::PixbufLoader->new();
        $loader->write( $image_data );
        $loader->close();
        $loader->get_pixbuf();
    };

So the next guy who looks at it doesn't have to wonder if $loader
is used anywhere later, and can see that the entire point of this
mumbo jumbo is to load something into $pixbuf.
Regards,
#Aristotle
</quote aristotle>
[download]