Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:76425
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain ajf.me designates 192.64.116.200 as permitted sender)
Content-Type: text/plain; charset=windows-1252
Mime-Version: 1.0 (Mac OS X Mail 7.3 \(1878.6\))
In-Reply-To: <53E8A447.5080303@gmail.com>
Date: Mon, 11 Aug 2014 12:26:16 +0100
Cc: "internals@lists.php.net" <internals@lists.php.net>
Content-Transfer-Encoding: quoted-printable
Message-ID: <2B6755F2-74F7-4D4D-99CC-319BFF78F2DF@ajf.me>
References: <D17EB1B8-7A7E-419B-AF55-0ADCAB4CFEB3@ajf.me> <53DF2BED.10103@sugarcrm.com> <06F3EF08-21B1-49AD-9D9D-5043C69AC1D8@ajf.me> <53DFE1FC.5040206@gmail.com> <BDDFD20F-1885-4106-BA7E-2A837C02ADF9@ajf.me> <BC228A66-1337-4A4B-8B7E-E2DBC641CD1B@fb.com> <B142BE0C-2E6F-413B-BAC2-19F17218AA05@ajf.me> <4c89fa2a-a4da-46a9-9872-6d2e5346fed1@email.android.com> <7AC9A96B-9B45-45B2-BBD1-1E61CC6075E8@ajf.me> <53E8A447.5080303@gmail.com>
To: Rowan Collins <rowan.collins@gmail.com>
Subject: Re: [PHP-DEV] [RFC] Closure::call and Function Referencing as Closures
From: ajf@ajf.me (Andrea Faulds)


On 11 Aug 2014, at 12:08, Rowan Collins <rowan.collins@gmail.com> wrote:

> I understand that &$foo->method won't work, but that's what I mean by =
pre-supposing the proposed syntax.
>=20
> If you had a different syntax, such as =A3function (I know it's not a =
sensible choice, but it's an example of something with no conflict), you =
*would* be able to implement =A3$foo->method. It would "do the same =
thing calls do", which is call a non-static method non-statically.
>=20
> So the question is, is it a) desirable and b) practical to come up =
with an implementation where that is possible?

Ah, I see what you=92re saying. It=92s certainly worth thinking about. & =
isn=92t really the nicest of syntaxes.

> Yep, that's my point - referencing a method without also referencing =
the class or object it's bound to is not something that usually happens =
in PHP, so it feels odd to make it the default behaviour here.
>=20
> Your proposal requires that to create a reference to a non-static =
method of the current object, you would do something like this:
>=20
> $foo =3D &static::foo;
> $foo->bindTo($this);
>=20
> We could imagine instead having a factory like Closure::wrap(callable =
$c), where the exact equivalent would be:
>=20
> $foo =3D Closure::wrap([static::class, 'foo']);
> $foo->bindTo($this);
>=20
> But you would never write it that way, because we have a way of =
specifying callables for instance methods, so you'd just write this:
>=20
> $foo =3D Closure::wrap([$this, 'foo']);
>=20
> You might for some reason want to re-bind that to a different object =
(although I'm not sure why, outside of things you'd expect to use =
Reflection for anyway), but the default would be referencing its =
existing binding.

I think Closure::bind(&FooBar::foo, =91FooBar=92); works well, but yes, =
I can see your point.

> In which case what do you think of my second suggestion, of adding =
methods to allow people to tell whether a closure is bound, bindable, =
etc? It makes it much easier to write defensive code if you can predict =
errors rather than catching them.

I=92m totally in favour of that. One thing I=92d like to see, if we =
introduce references for functions and classes and so on, is more =
reflection-like stuff. I think it=92s a bit silly to have to instantiate =
reflection classes all the time. Given that closures are already =
objects, I see no reason why you should need to use ReflectionFunction =
on them. Rather, we should provide all this information on the object =
directly.

--
Andrea Faulds
http://ajf.me/