Re^2: Safer monkey-patching

[tobyink]

Indeed, this problem is not really solvable. However, my solution does offer a partial way forward.

Let's imagine that My::AwesomeError and Your::GroovyError both superclass Example::Error, and both add an asplode method.

We can then do:

my $err = Example::Error->new('Heat death of universe detected!');
$err->My::AwesomeError::asplode();
$err->Your::GroovyError::asplode();
$err->asplode();  # Russian roulette

With true monkey patching, where My::AwesomeError and Your::GroovyError each defined the asplode method directly in the Example::Error namespace, we wouldn't get the option to do the above.

[moritz]

We can then do:

my $err = Example::Error->new('Heat death of universe detected!');
$err->My::AwesomeError::asplode(); $err->Your::GroovyError::asplode();
$err->asplode(); # Russian roulette

Of course you can, but it's not really a benefit. As you wrote in the OP, monkey patching is often done to work around limitations in existing code that you don't want to patch. But writing $obj->Class::method() requires the author of that code to be aware of the issue, and when he is, he can simply write Class::method($obj) without requiring any monkey patching.

I do find your idea cute, but it really doesn't solve the "hard" problem, and the small problems it does solve seem to be either artifically constructed, or non-issues.