Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:83655
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.212.175 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <CAF+90c-JngNyAvz5m2MP67qH4JQuXGhkYtJoatWiXVAkjAsMcA@mail.gmail.com>
References: <CAF+90c-JngNyAvz5m2MP67qH4JQuXGhkYtJoatWiXVAkjAsMcA@mail.gmail.com>
Date: Tue, 24 Feb 2015 12:37:47 +0100
Message-ID: <CAF+90c9XgS6D7_vZLaA6rhg=Z-tGQzX0Pa0_HN93eZFTv4nPYA@mail.gmail.com>
To: PHP internals <internals@lists.php.net>
Content-Type: multipart/alternative; boundary=047d7b3a834cb03126050fd3f206
Subject: Re: Mark some functions with PREFER_REF
From: nikita.ppv@gmail.com (Nikita Popov)

--047d7b3a834cb03126050fd3f206
Content-Type: text/plain; charset=UTF-8

On Mon, Feb 23, 2015 at 10:36 PM, Nikita Popov <nikita.ppv@gmail.com> wrote:

> Hi internals!
>
> With regard to my E_STRICT RFC [1] Yasuo and another off-the-record mail
> expressed a concern about the "Only variables should be passed by
> reference" strict standards warning.
>
> This notices makes sense in a lot of places, e.g.
>
>     array_push(getArray(), 1, 2, 3);
>
> is totally pointless code if getArray() doesn't return a reference.
> array_push will just modify a temporary array in this case.
>
> However other usages like
>
>     $top = array_pop(getArray());
>
> are less clear: array_pop() both has a side-effect (removing the top
> element of an array) and a meaningful return value (the top element of the
> array). As PHP does not have a dedicated function for getting the
> first/last element of an array, many people use code like this (which
> currently throws E_STRICT) instead.
>
> I'd like to mark a few such functions with PREFER_REF (i.e. allow passing
> both references and values). In particular array_pop() and array_shift().
>
> Does anybody have a problem with us doing this? Are there other functions
> which should use PREFER_REF?
>
> The motivation behind this is to make sure we only throw this notice in
> cases where the code is actually buggy/nonsensical instead of just making
> people ignore it altogether.
>

A few people have pointed out OTR that the same applies to reset() and
end(). Both can be used to get the first/last element (and change the IAP
as a side effect). So I'd add PREFER_REF to those as well. (Interestingly
key() and current() already use PREFR_REF even though they don't modify the
array at all, probably this is a performance optimization.)

Nikita

--047d7b3a834cb03126050fd3f206--