Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:94892
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.161.177 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <CANaX5ne1BHOQ4Q8OUPWx-=Sg43zd6L54HZ7DbVzUZniTk9hkKw@mail.gmail.com>
References: <CANaX5ne1BHOQ4Q8OUPWx-=Sg43zd6L54HZ7DbVzUZniTk9hkKw@mail.gmail.com>
Date: Sun, 7 Aug 2016 09:58:24 +0200
Message-ID: <CAF+90c8ZBy4OE-hg=f334e8QyL8pbKtq_AchTimRyr4ROn_vdw@mail.gmail.com>
To: Davey Shafik <me@daveyshafik.com>
Cc: "internals@lists.php.net" <internals@lists.php.net>
Content-Type: multipart/alternative; boundary=94eb2c07e31e0791cf053976aa96
Subject: Re: [PHP-DEV] Undocumented Nullable Return Behavior
From: nikita.ppv@gmail.com (Nikita Popov)

--94eb2c07e31e0791cf053976aa96
Content-Type: text/plain; charset=UTF-8

On Sun, Aug 7, 2016 at 9:07 AM, Davey Shafik <me@daveyshafik.com> wrote:

> Hi internals,
>
> In the nullable types we do not allow implicit return null/return with no
> value to fulfill a nullable type.
>
> - No return will result in a TypeError when executed
> - Return with no value will be an actual fatal error, which I believe is
> due to the fact it's not found at runtime, should be a TypeError or
> ParseError if possible?
>

It's a compile time error. Compile time errors are currently fatal errors
for technical reasons.


> - Return explicit null is fine (as expected)
>
> See: https://3v4l.org/h6IYF
>
> I believe the first two are contrary to PHP's "null is always returned if
> nothing else is".
>
> OTOH, it makes sense as it's the entire reasoning behind return void vs
> return null.
>

Indeed. The current behavior is consistent with the (explicit decision
from) the void RFC.

Additionally, the current behavior is consistent with the behavior of the
implementation at the time the RFC was voted, with the only difference that
I have made this a compile-time error after the fact (because I think if we
know that something will always result in a runtime error, we should try to
detect it at compile-time already, and because the void error is detected
at compile-time as well).

Nikita


> This behavior is completely undocumented in the RFC.
>
> It's definitely something we should make an explicit decision on for the
> language spec if nothing else.
>
> We should also make any decisions on this prior to beta3 in < 2 weeks.
>
> Thoughts?
>
> - Davey
>

--94eb2c07e31e0791cf053976aa96--