Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:99250
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: softfail (pb1.pair.com: domain golemon.com does not designate 74.125.82.44 as permitted sender)
MIME-Version: 1.0
Sender: php@golemon.com
In-Reply-To: <CAFMT4Nqyhv8-XaC7L8khHjGKRMW677AWeH8red2ogmCEYV-mSg@mail.gmail.com>
References: <CAFMT4Nrt7XPMnELoAGLu=sSRCTCtzPCTaSE8HJjKfHjDXOXU_A@mail.gmail.com>
 <97bb5051-6413-b78a-e981-ad8858caec4c@gmail.com> <8d4ee559-1ad2-4d65-ae4f-207e59f056e3@Spark>
 <61DC1C28-CAF2-45F4-8D9F-A66BEA2E90B1@gmail.com> <3942113e-f83a-444f-bbc2-462c70d67c65@Spark>
 <20db9b97-5313-057d-db56-920ccb758558@telia.com> <CAFMT4Noi3s1eB9Og-c3P=wnz646G=eeSmCASVF1q_JM-SVvZKA@mail.gmail.com>
 <877970f0-b22d-a8b9-9ad4-6fc9139a9b34@telia.com> <CAFMT4Nqyhv8-XaC7L8khHjGKRMW677AWeH8red2ogmCEYV-mSg@mail.gmail.com>
Date: Mon, 29 May 2017 23:15:39 -0400
Message-ID: <CAESVnVpmV+M-Xw+=ePduZDDJ==HwkOyTT_+SvxcbTYm1pozxog@mail.gmail.com>
To: Levi Morrison <levim@php.net>
Cc: =?UTF-8?Q?Bj=C3=B6rn_Larsson?= <bjorn.x.larsson@telia.com>, 
	internals <internals@lists.php.net>
Content-Type: text/plain; charset="UTF-8"
Subject: Re: [PHP-DEV] [RFC][Discuss] Arrow Functions
From: pollita@php.net (Sara Golemon)

On Mon, May 29, 2017 at 10:48 PM, Levi Morrison <levim@php.net> wrote:
> Based on the discussion there are a few different syntax choices that
> we probably just need to vote between. It's a feature that people seem
> to want but everyone seems to prefer a different syntax choice.
>
>   1. fn(params) => expr
>   2. function(params) => expr
>
>   3. (params) ==> expr
>   4. (params) => expr
>
TL;DR - I'll be happy with whatever the group can agree on, with a
personal ordering of: 1 > 3 > 2.

I excluded #4 because I just don't see how you disambiguate this: [
($x) => $x ] . It's already valid syntax: https://3v4l.org/NedDu .
With zero or 2+ args it's disambiguable, but one arg has to be
possible or it's a no-go.

> This is why everyone is so divisive. All options have drawbacks.
> Additionally some people don't like binding by value and would prefer
> ref, and others really would be against by-ref.
>
Again, I'll accept the group's opinion because I don't care enough to
block it, but I'm registering my preference for the record, as by-ref.

> Which brings me to an option I don't think was ever discussed on list:
>
>      [](params) => expr     // binds no values
>      [=](params) => expr    // binds by value
>      [&](params) => expr    // binds by reference
>
As a C++ developer, I have no inherent problems with this **in
theory**, but I agree with your statement that it's a bit symbolic
soupish.  I also suspect that we'll run into grammar ambiguities.
Imagine:   [ []($x) => $x ].  This is current legal syntax (at the
point of compilation) because it *could* be interpreted as an array
with a single element (value $x) who's key is the result of calling a
function defined by an empty array passing a single arg ($x).  It's
invalid at runtime (See: https://3v4l.org/IcK3B ), but that means
special casing our way around the ambiguities.

Further, if we're going to follow the C++ model, we'd be well suited
to allow explicit per-var by-val/by-ref syntax, e.g.:

  [$foo, $bar]() => $foo * $bar

Again, this is legal syntax to the parser, but now it's even
potentially current legal syntax to the runtime (assuming
is_callable([$foo, $bar]).

So yeah, probably doesn't fit PHP, but good thinking outside the box. :D

-Sara