Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:56803
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.215.170 as permitted sender)
MIME-Version: 1.0
Sender: ekneuss@gmail.com
In-Reply-To: <CAMUwpuR0gw3vkGeEFp9pCtpmQ075TLmLSvJDSVRRc+40_KBGEQ@mail.gmail.com>
References: <CAMUwpuR0gw3vkGeEFp9pCtpmQ075TLmLSvJDSVRRc+40_KBGEQ@mail.gmail.com>
Date: Tue, 6 Dec 2011 15:01:25 +0100
Message-ID: <CACQesk5eHCB06dOv3K9om_2xo90iAZRtm_tWHAUbposEnOmkJA@mail.gmail.com>
To: jpauli <jpauli@php.net>
Cc: internals@lists.php.net
Content-Type: multipart/alternative; boundary=000e0cd1e24cf9ec5504b36cde2d
Subject: Re: [PHP-DEV] One more Liskov Subs. and parameter checking notice
From: colder@php.net (Etienne Kneuss)

--000e0cd1e24cf9ec5504b36cde2d
Content-Type: text/plain; charset=UTF-8

Hi,

On Tue, Dec 6, 2011 at 14:54, jpauli <jpauli@php.net> wrote:

> Greetings PHP Intern@ls :)
>
> There is a PHP behavior I disagree with regarding OO model.
> It's been here for a long time (5.0 ??, at least 5.2, 5.3 and 5.4) , here
> it is :
>
> <?php
> interface Iface1 { }
> interface Iface2 extends Iface1 { }
>
> class Foo {
>    public function bar(Iface1 $arg) { }
> }
>
> class Foo2 extends Foo {
>    public function bar(Iface2 $arg) { }
> }
> ?>
> Strict Standards: Declaration of Foo2::bar() should be compatible with that
> of Foo::bar()
>
>
> I find this wrong.
> Liskov (and the error message) says we must stay "compatible" in our
> inheritence.
> The fact is that we are in the code above :
> - A uses IfaceA
> - B extends A
> - B should be able to overwrite A's method typing their IfaceA params on
> IfaceB, if IfaceB extends IfaceA (and is thus compatible with it).


> Thoughts ?
>

No, that makes B incompatible with A.

Consider calling $obj->bar($obj) where obj implements IFaceA and not
IfaceB, if $obj is of class A, this is ok, if $obj is of class B, this is
invalid.

In other terms, arguments' types should be contra-variant. Currently PHP
expects them to be invariant, but there is already an RFC discussing how to
extend it:

https://wiki.php.net/rfc/prototype_checks

Best,


>
> Julien.Pauli
>



-- 
Etienne Kneuss
http://www.colder.ch

--000e0cd1e24cf9ec5504b36cde2d--