Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:65698
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.210.49 as permitted sender)
Content-Type: text/plain; charset=iso-8859-1
Mime-Version: 1.0 (Mac OS X Mail 6.2 \(1499\))
In-Reply-To: <CAAyV7nFV6bT822U+BkydhrVWZ8nq8ROPZXq4UZpKy12_UG6fcw@mail.gmail.com>
Date: Wed, 6 Feb 2013 09:09:42 -0800
Cc: Etienne Kneuss <colder@php.net>,
 Mike Willbanks <pencap@gmail.com>,
 PHP Internals <internals@lists.php.net>
Content-Transfer-Encoding: quoted-printable
Message-ID: <BDF0960A-4B7C-453F-ADA3-99991402FD46@gmail.com>
References: <CALrkBjZ1Jb9k1=b8uyqbmaMojw06vSaAZCMfDU3LnXGP2cWEMw@mail.gmail.com> <CACQesk6dr8+0CkAgZOca5L=PipdPbt8sSFf_qvUn7r5pjqXnYg@mail.gmail.com> <CAAyV7nFV6bT822U+BkydhrVWZ8nq8ROPZXq4UZpKy12_UG6fcw@mail.gmail.com>
To: Anthony Ferrara <ircmaxell@gmail.com>
Subject: Re: [PHP-DEV] ArrayAccess/ArrayObject return by reference
From: sommertm@gmail.com (Tyler Sommer)

How would the interface enforce returning a reference?


On Feb 6, 2013, at 8:47 AM, Anthony Ferrara <ircmaxell@gmail.com> wrote:

> Perhaps there's another way out of this. A simple way would be to =
introduce
> an ArrayAccessReference interface in core that adds the references to =
the
> getters/setters...
>=20
> It's perhaps not the cleanest, but it solves the BC issues...
>=20
> Anthony
>=20
>=20
> On Wed, Feb 6, 2013 at 11:42 AM, Etienne Kneuss <colder@php.net> =
wrote:
>=20
>> I assume it would be possible technically but might break (at least =
by
>> issuing E_STRICT) a lot of code if we forced ArrayObject::offsetGet =
to
>> return a reference.
>>=20
>> Think of all the subclasses that extend ArrayObject who currently do =
not do
>> that?
>>=20
>> Other than that, returning a ref where it previously didn't can have =
all
>> kinds of undesirable and hard-to-track consequences.
>>=20
>> Best,
>>=20
>>=20
>> On Wed, Feb 6, 2013 at 4:27 PM, Mike Willbanks <pencap@gmail.com> =
wrote:
>>=20
>>> Looking at: =
http://lxr.php.net/xref/PHP_5_5/Zend/zend_interfaces.c#538it
>>> seems that ArrayAccess at the moment can not be returned by a =
reference.
>>> I'm wondering if there was a technical reason behind this or if it =
is
>> now
>>> a BC reason?
>>>=20
>>> Anyhow; I was attempting to dig through the source code a bit more =
last
>>> night to see why ArrayObject could not overload the function =
declaration
>> of
>>> offsetGet to force a return by reference aka: function
>> &offsetGet($key)...
>>> which works now for ArrayAccess but not for ArrayObject.  I believe =
it
>> has
>>> to deal with ArrayObject inheriting ArrayAccess?  Is there a way to =
allow
>>> ArrayObject to change the function declaration in this way?  My PHP
>>> internals skills are not the best which is the reason for the =
question.
>>>=20
>>> Anyhow; justification wise: in userland this leads to a lot of wtf
>> factor.
>>> It really comes down to having to provide our own implementation of
>>> ArrayObject by extending several different areas including =
ArrayAccess so
>>> that references can be returned so multi-dimensional arrays can be
>> properly
>>> unset aka:
>>> $ar =3D new ArrayObject(array('foo' =3D> array('bar' =3D> =
array('baz' =3D>
>>> 'foo'))));
>>> unset($ar['foo']['bar']['baz']);
>>>=20
>>> Regards,
>>>=20
>>> Mike
>>>=20
>>=20
>>=20
>>=20
>> --
>> Etienne Kneuss
>> http://www.colder.ch
>>=20