Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:102830
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.221.67 as permitted sender)
Date: Sun, 15 Jul 2018 17:31:10 +0100
User-Agent: K-9 Mail for Android
In-Reply-To: <CAFMT4NoZEivQtoueJ4B3aZvaJNGwjafcGucLo7xyp5qyNK-m5Q@mail.gmail.com>
References: <8916EC21-D368-40F8-9ABD-CE0C04A73539@gmail.com> <CAFMT4Np3we41HEYz3V4-JLGdxr9ToOMh9k=BpZR_idJTfSqcig@mail.gmail.com> <6E843F96-7815-4525-8AB1-D9D700B0D031@gmail.com> <CAFMT4NoZEivQtoueJ4B3aZvaJNGwjafcGucLo7xyp5qyNK-m5Q@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain;
 charset=utf-8
Content-Transfer-Encoding: quoted-printable
CC: internals <internals@lists.php.net>
Message-ID: <DE80349B-E92D-4E40-A606-86EF98CDDAE7@gmail.com>
Subject: Re: [PHP-DEV] Non-nullable properties
From: rowan.collins@gmail.com (Rowan Collins)

On 14 July 2018 19:29:23 BST, Levi Morrison <levim@php=2Enet> wrote:
>Whether it's a constructor, a factory method, or just a block of code
>near the `new` site: something has the responsibility for
>initialization=2E

Indeed, and my suggestion is to formalise that responsibility, so that mis=
takes can be detected closer to their source - which is, after all, the who=
le point of type constraints=2E


> Once initialized the object provides guarantees unless
>you `unset` the property=2E

But since you have no way of knowing if it was initialized or not, this is=
n't a guarantee you can actually base code on=2E


> It also provides the guarantee that if it is
>set then it will always be of the outlined type; this is a much
>stronger guarantee than we have today, which is=2E=2E=2E nothing=2E

I'm not saying we should block the whole feature (apologies if that wasn't=
 clear), so that's not the appropriate comparison=2E The comparison I'm int=
erested in is between writing=2E=2E=2E

/**
 * Will be initialised in the constructor and should not be unset=2E
 */
public Foo $foo;

=2E=2E=2E and =2E=2E=2E

/**
 * Will be initialised in the constructor and should not be set to null=2E
 */
public ?Foo $foo=3Dnull;

In both cases, the intent is documented, but not enforced=2E


If we allow the first version, with no actual checks on the initialisation=
, it will be very hard to introduce a stricter version later=2E If we only =
allow the second version in the initial implementation, we lose very little=
=2E

The rule could be very simple: typed properties must specify a valid defau=
lt value=2E That means object hints will have to be nullable, with a defaul=
t of null, and there is no need to invent a new kind of "null reference"=2E

Regards,

--=20
Rowan Collins
[IMSoP]