Thank you !

I have one last small question (shall i ask it in another node ?) :
When i load my pages, it takes some time. Most of the time, i don't need to wait for the images to be loaded in the page, just the HTML content, and sometimes the DOM.
I found an imperfect solution, and am looking for the a better one.
At the moment i use

synchronize => 0,
[download]

I think i need to use the "events" parameter of the WWW:Mechanize::Firefox contructor. For example

events => ['DOMContentLoaded'],
[download]

I am sorry, i have read on this subject several time in the past, and have searched again today. I am making you repeat. The best solution i read seems to be

events => ['DOMContentLoaded'],
[download]

I'm not aware of a faster method than waiting for the DOMContentLoaded event. You can speed things up if you only need to wait for certain elements of the page becoming available, but if you want the DOM to be complete, I know of no better event to wait for.

Wait for one particular element, yes that could be it.
* time passes and i search *
I found it !
I had seen it several weeks ago, but had not understood at the time, and then forgotten.
In the cookbook, HERE.

Here is my version, and it doesn't work. When i call the WWW::Mechanize::Firefox::is_visible() subroutine, it stops in the middle and says "10 elements found for '//a@class="kl"' at test_events_synchronize.pl line 51."
Apparently this happens in find_link_dom(), and it highlights many elements in the window. What am i doing wrong? Here is my code :

#!/usr/bin/perl 

use 5.012003;    # Perl version requirement

use strict;
use warnings;    #comment before releasing

use autodie              2.10;
use Carp                 1.25;
use Perl6::Say           0.16;
use Readonly             1.03;
use Scalar::Util                qw( dualvar );
#
#####

use WWW::Mechanize::Firefox 0.59;
use Time::HiRes             1.9721  qw( time );

# where is my Firefox executable ? 
Readonly my $FIREFOX_FOLDER 
    => 'C:\Program Files\Mozilla Firefox\firefox.exe';
#
######

# URL address of the two pages that we want to open in two tabs
Readonly my $URL => 'https://www.google.ca/search?tbm=isch&hl=en&sourc
+e=hp&biw=1400&bih=866&q=big+images&gbv=2&oq=big+images&aq=f&aqi=g10&a
+ql=&gs_sm=3&gs_upl=1564l3028l0l3402l10l10l0l0l0l0l142l1032l4.6l10l0&g
+s_l=img.3..0l10.1564l3028l0l3403l10l10l0l0l0l0l142l1032l4j6l10l0';

# start time
my $t_start = time;

# start browser 1
my $browser = WWW::Mechanize::Firefox->new(
    # launch Firefox if it is not already running
    launch => $FIREFOX_FOLDER,
    
    # events to wait for, before continuing after get()ing a page
#    events => ['DOMContentLoaded'], 
    synchronize => 0,
);
croak 'Could not initialize the browser' if !$browser->success();
say '# Browser is now ready to go';

# open page
$browser->get( $URL );
croak 'Could not open the page' if !$browser->success();
say '# Page opened';

my $test_to_execute = 1;

if( $test_to_execute == 1 ) { 
    # wait until one particular element appears 
    my $retries = 10;
    while ($retries-- 
           && ! $browser->is_visible( xpath => '//a[@class="kl"]' )) {
        sleep 1;
    };
    die "Timeout" unless $retries;
}
elsif(  $test_to_execute == 2 ) { 
    my $test_is_visible = $browser->is_visible( xpath => '//a[@class="
+kl"]' );
    say $test_is_visible;
}

# print time passed
my $t_passed = dualvar time - $t_start, 'time passed';
say "$t_passed = " . ($t_passed+0);

# stop the program, so i can check which pages have been opened
say '# done !';
my $stop = <ARGV>
[download]

PS : if is set

$test_to_execute = 2
[download]

i get exactly the same result.