«

X-No-archive: yes

This is perl, v5.6.1 built for MSWin32-x86-multi-thread
-----
package Rational;

use overload "0+" => \&float;

# There is a subroutine called float() that, well, returns a float.
------
use Rational;
..
..
..
#$fraction holds an object of type Rational
my $fsq = sqrt($fraction);
------------
perl complains:
Operation sqrt: no method found, argument in overloaded package Rational at ...

Since sqrt() takes a numeric argument, I expected the overloading of "0+"
to force float() to be called on the object $fraction, but it doesn't work.
The code "my $fsq = sqrt($fraction->float());" compiles OK.
Why isn't numification working in this case?

(Unknown Poster) wrote:

: X-No-archive: yes

Why? Yours is a good question.

: package Rational;
:
: use overload "0+" => \&float;
:
: # There is a subroutine called float() that, well, returns a float.
: ------
: use Rational;
: #$fraction holds an object of type Rational
: my $fsq = sqrt($fraction);
: ------------
: perl complains:
: Operation sqrt: no method found, argument in overloaded package Rational at ...
:
: Since sqrt() takes a numeric argument, I expected the overloading of "0+"
: to force float() to be called on the object $fraction, but it doesn't work.

Because sqrt() is itself an overloadable operator.

You can either let the overload pragma handle this on its own:

use overload
"0+" => \&float,
fallback => 1;


or, probably better, go ahead and overload sqrt:

use overload
"0+" => \&float,
"sqrt" => sub { sqrt($_[0]->float) };