Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:46100
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 74.125.78.27 as permitted sender)
DomainKey-Status: bad
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:sender:from:date:x-google-sender-auth:message-id
         :subject:to:cc:content-type;
        b=WxJtZu8B4bFUXLLDcM6gEXEIF55G9TOKwJ7taH4Bwdkz69ioC/OVGdcCZhRwj1rjWY
         aII/UR7ggU7BEBHpOmUUExzFB9Rapz5i7txPXoaKNMfsInUKPG9+71zaG7JPjMWZkO3b
         MdnTfd33CjQaMDMPH3Z7DiNyTbGuW3xQeDHHs=
MIME-Version: 1.0
Sender: ametaireau@gmail.com
Date: Thu, 19 Nov 2009 12:50:42 +0100
Message-ID: <2229b0db0911190350j6d0bde4cgcb57675778b55d0b@mail.gmail.com>
To: internals@lists.php.net
Cc: fred@spiral-project.org
Content-Type: multipart/alternative; boundary=00151744790e3561660478b7f9e3
Subject: RFC: Custom Factories (SPL)
From: alexis@spiral-project.org (=?UTF-8?Q?Alexis_M=C3=A9taireau?=)

--00151744790e3561660478b7f9e3
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Hello here!

I'm a big fan of IoC too, and I think that PHP devs really need to have a
look into Inversion of control. Adding this feature to PHP can be useful fo=
r
providing a simple way to detect and manage the object creation process.

The "spl_factory_register" function allows to specify a manner to catch whe=
n
a "new" occurs, and to add an alternative way to manage the creation of
objects, by calling custom factories instead of using the built in "new"
process.

In fact, this method allow us to branch our objects managers (or dependency
injection containers) on classes that don't implement inversion of control
(like the widely used singleton pattern). This is huge ! However  it's
important to note that* this behavior can encourage some people to continue
writing hard coupled classes*, and use don't us or think about inversion of
control, letting the (service/object) container do the stuff for us. It can
be really *dangerous*, and maybe attack the problem by the wrong side ?

On a pragmatic point of view, *this is an actual need*, for sure. At least
Richard and me have encountered theses problems (and I think many other
people). This "spl_factory_register" would solve them in an elegant way.

About what Eloy Bote said, I think that traits and IoC are two different
concepts, both talking about code re-usability and good practices, and both
needed.

And, about the signature of the function itself, why not use a callback lik=
e
in "spl_autoload_register" function ?

Best regards
--=20
Alexis M=C3=A9taireau
06 20 15 55 24 | 35 rue L=C3=A9o Lagrange, 31400 Toulouse |
http://www.notmyidea.org

--00151744790e3561660478b7f9e3--