Maybe it doesn't make sense to call <<Paste>> on a top level, but then why does it work with <Control-v>. I'm willing to bet these two events are bound to the exact same Paste subroutine within the Tk::Text code (but I haven't actually looked, but Tk::TableMatrix is that way).

I think what Paste is supposed to do isn't really relevant. You could rebind <<Paste>> and <Contol-v> to a subroutine that does nothing more than print "Hello World" to the console, or one could make up a new virtual event and use some unused key sequence event and bind them to a subroutine that just prints "hi". The point is, the window manager seems to treat a virtual event and a physical event in distinctly different ways. I'm just not clear on exactly how different and for what reason.

My theory at this point is that when you call a virtual event with $w->eventGenerate(<<virtualevent>>), then only the callback for the binding to "virtualevent" that matches widget w$ and it's parent widgets will be called. But when a physcial event is called by $w->eventGenerate(<physicalvent>) then the callback for all bindings to "physicalevent" matching any active widgets contained in $w's toplevel will be called.

Here are some examples. In the first two examples, since we are generating physical events, all three bindings get called no matter whether the eventGenerate is called by the toplevel $mw, or the text widget $t.

Example 1

#!/usr/bin/perl -w

use strict;
use Tk;
my $mw = MainWindow->new;

$mw->Button(-text => "Paste", -command => \&DoPaste)->pack();
my $t = $mw->Text(-width => 30, -height => 5)->pack();
$t->focus;

$mw->bind('<Control-t>' => \&Print1);
$mw->bind('Tk::Text', '<Control-t>' => \&Print2);
$t->bind('<Control-t>' => \&Print3);

sub DoPaste
{
    $mw->eventGenerate('<Control-t>');
}

sub Print1
{
    my $tw = shift;
    print "########\n";
    print $tw, "\t", ref($tw), "\n";
    print "1\n";
    print "###########\n";
}

sub Print2
{
    my $tw = shift;
    print "########\n";
    print $tw, "\t", ref($tw), "\n";
    print "222222222\n";
    print "###########\n";
}

sub Print3
{
    my $tw = shift;
    print "########\n";
    print $tw, "\t", ref($tw), "\n";
    print "33333 wheeee!\n";
    print "###########\n";
}


Tk->MainLoop;
[download]

Example 2 just change $mw->eventGenerate('<Control-t>'); in Example 1 to $t->eventGenerate('<Control-t>');

Now in Examples 2 and 3 we have replaced Control-t with an arbitrary virtual event name. Otherwise examples 2 and 3 are identical to 1 and 2, respectively. And yet Example 3 only calls the one callback, though Example 4 generates all of them just like Examples 1 and 2..

Example 3

#!/usr/bin/perl -w

use strict;
use Tk;
my $mw = MainWindow->new;

$mw->Button(-text => "Paste", -command => \&DoPaste)->pack();
my $t = $mw->Text(-width => 30, -height => 5)->pack();
$t->focus;

$mw->bind('<<Jayhawk>>' => \&Print1);
$mw->bind('Tk::Text', '<<Jayhawk>>' => \&Print2);
$t->bind('<<Jayhawk>>' => \&Print3);

sub DoPaste
{
    $mw->eventGenerate('<<Jayhawk>>');
}

sub Print1
{
    my $tw = shift;
    print "########\n";
    print $tw, "\t", ref($tw), "\n";
    print "1\n";
    print "###########\n";
}

sub Print2
{
    my $tw = shift;
    print "########\n";
    print $tw, "\t", ref($tw), "\n";
    print "222222222\n";
    print "###########\n";
}

sub Print3
{
    my $tw = shift;
    print "########\n";
    print $tw, "\t", ref($tw), "\n";
    print "33333 wheeee!\n";
    print "###########\n";
}


Tk->MainLoop;
[download]

Example 4 just change $mw->eventGenerate('<<Jayhawk>>'); in Example 1 to $t->eventGenerate('<<Jayhawk>>');

maybe this all much ado about nothing even if my theory is correct, I don't know, it just seems weird, I would have expected virtual events and physcial events to work the same way, and I've never seen anything in the documentation for Tk::bind or Tk::event or Tk::callback that would indicate they shouldn't act the same.

But they clearly do, right? Or am I crazy? Or did I miss something in the documentation that explains this feathure?