Hey,
Bug 48215 was a BC break from the previous 5.2 behaviour, it stemmed
from a change for bug #39127.
class a { function a($arg='') { echo $arg; } }
class b extends a {}
$b = new b;
$b->b('foo');
$b->__construct('foo');
This prints out an error in 5.3 which is fine, the unexpected change
was that if a __construct() method was actually defined as well as a
method that matched the class name it would alias the parent method to
the constructor. The test is attached in the patch.
Just sending off to internals first since we're so late in the RC stage.
Scott
Hi,
Hey,
Bug 48215 was a BC break from the previous 5.2 behaviour, it stemmed
from a change for bug #39127.
this should be fixed in 5.3, for me the patch looks fine, and should be
committed. While I'd be in favor on a short review by somebody else
(Dmitry?)
Thanks,
johannes
class a { function a($arg='') { echo $arg; } }
class b extends a {}$b = new b;
$b->b('foo');
$b->__construct('foo');This prints out an error in 5.3 which is fine, the unexpected change
was that if a __construct() method was actually defined as well as a
method that matched the class name it would alias the parent method to
the constructor. The test is attached in the patch.Just sending off to internals first since we're so late in the RC stage.
Scott
The patch looks fine to me.
Ilia Alshanetsky
Hey,
Bug 48215 was a BC break from the previous 5.2 behaviour, it stemmed
from a change for bug #39127.class a { function a($arg='') { echo $arg; } }
class b extends a {}$b = new b;
$b->b('foo');
$b->__construct('foo');This prints out an error in 5.3 which is fine, the unexpected change
was that if a __construct() method was actually defined as well as a
method that matched the class name it would alias the parent method
to the constructor. The test is attached in the patch.Just sending off to internals first since we're so late in the RC
stage.Scott
<bug48215.patch.txt