Bug #78372 [Dup]: Calling __set magic method on ReflectionProperty::setValue()

From: Date: Mon, 05 Aug 2019 09:18:05 +0000
Subject: Bug #78372 [Dup]: Calling __set magic method on ReflectionProperty::setValue()
References: 1  Groups: php.bugs 
Request: Send a blank email to php-bugs+get-222089@lists.php.net to get a copy of this message
Edit report at https://bugs.php.net/bug.php?id=78372&edit=1

 ID:                 78372
 Updated by:         cmb@php.net
 Reported by:        i dot spyric at gmail dot com
 Summary:            Calling __set magic method on
                     ReflectionProperty::setValue()
 Status:             Duplicate
 Type:               Bug
 Package:            Scripting Engine problem
 Operating System:   Debian
 PHP Version:        7.4.0beta1
 Block user comment: N
 Private report:     N

 New Comment:

See <https://wiki.php.net/rfc/typed_properties_v2#property_initialization>.


Previous Comments:
------------------------------------------------------------------------
[2019-08-05 08:50:17] requinix@php.net

Then really this is a duplicate: the typed properties + __set problem, this time using reflection.

The awkwardness is specifically about the property being uninitialized; once set, or when
specifically unset(), __get/etc. act as expected. So it's looking like a class cannot
effectively use uninitialized typed properties as well as __get/etc, and that the best option will
be to always initialize them with something - even if 0 or "" or false. (I myself already
do this with untyped properties.)

In fact this feels like it'll be such a surprise that I would say maybe typed properties *must*
be initialized, in the declaration itself:
  private string $test = "";
I don't see this option mentioned in the RFC and I don't remember if it came up during the
RFC discussion. Rationale is that every property should always end up either with some real value or
as null to represent a lack of value; for the former this default won't make much of a
difference since it will be overwritten, and for the latter the type should have been ?nullable and
so could be implicitly initialized by PHP (to null).

------------------------------------------------------------------------
[2019-08-05 07:55:14] nikic@php.net

While this is technically "expected", I do think that we may need to do something about
this, though I'm not sure what. See also bug #78226.

------------------------------------------------------------------------
[2019-08-05 07:20:45] requinix@php.net

The RFC states that __get will be invoked upon reading an uninitialized typed property. Logically,
__set should be invoked when writing an uninitialized typed property.

This behavior matches how an unset() untyped property works, even when using reflection. https://3v4l.org/Gs1XL

------------------------------------------------------------------------
[2019-08-05 07:10:32] i dot spyric at gmail dot com

Description:
------------
When you use setting value via reflection on uninitialized property __set magic method will be
called. 

If you remove type of property or set default value, __set method will not be called.

Is it a bug? Because in RFC I have not seen anything about this behavior

Test script:
---------------
class Example {
    private string $test;

    public function __set($name, $value)
    {
        echo "$name=$value";
    }
}

$t = new Example();

$property = new ReflectionProperty($t, 'test');
$property->setAccessible(true);
$property->setValue($t, 'foo');

Expected result:
----------------
No __set is called

Actual result:
--------------
__set is called


------------------------------------------------------------------------



--
Edit this bug report at https://bugs.php.net/bug.php?id=78372&edit=1


Thread (5 messages)

« previous php.bugs (#222089) next »