Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:128479
Feedback-ID: id5114917:Fastmail
Date: Fri, 15 Aug 2025 18:04:58 +0100
To: PHP internals <internals@lists.php.net>
Subject: Re: [PHP-DEV] Split as an alias to explode
User-Agent: K-9 Mail for Android
In-Reply-To: <CAEwRQKZZFpGxbJ7RUJm1qzEnZwxq31v4Tz4RWK-GbNpYssmQhw@mail.gmail.com>
References: <CAEwRQKZZFpGxbJ7RUJm1qzEnZwxq31v4Tz4RWK-GbNpYssmQhw@mail.gmail.com>
Message-ID: <88658822-7BB5-4407-B9E1-3A617DAE7581@rwec.co.uk>
Precedence: list
MIME-Version: 1.0
Content-Type: text/plain;
 charset=utf-8
Content-Transfer-Encoding: quoted-printable
From: imsop.php@rwec.co.uk ("Rowan Tommins [IMSoP]")

On 15 August 2025 16:53:52 BST, Vinicius Dias <carlosv775@gmail=2Ecom> wrot=
e:
>Hello, internal folks=2E
>
>We have `join` as an alias for `implode` so I think it makes sense to
>have `split` as an alias to `explode`, doesn't it?


There's some history here that won't be obvious to anyone who has been usi=
ng PHP for less than a decade=2E=20

In PHP 4 and 5 (and maybe even earlier) PHP had separate functions called =
explode() and split()=2E One split the string by an exact delimiter, the ot=
her split it by a regular expression pattern=2E For many use cases, these w=
ere interchangeable, because a pattern with no special characters in was ma=
tched literally; but in other cases, it caused a lot of confusion when spli=
t() gave unexpected results=2E=20

The entire "Posix-Compatible Regular Expressions" extension (ext/ereg) was=
 deprecated in PHP 5=2E3, and removed in 7=2E0=2E That included the split()=
 function, leaving us with only explode()=2E

At that point, making split() an alias of explode() would have been possib=
le, but unhelpful: rather than failing with "no such function", existing co=
de would have subtly changed behaviour=2E=20

However, we have now had eight full releases with no built-in function cal=
led split(), so the chance of somebody upgrading in one go from PHP 5=2E6 t=
o PHP 8=2E5 and missing the change seems very low=2E If we wanted to be rea=
lly conservative, we could plan the alias for 9=2E0 so there were two clear=
 major versions between removal and reuse of the name=2E

I believe straightforward aliases like this are a single line in a stub fi=
le, unless there's any compiler optimisations directly referencing the name=
=2E

As such, I would broadly support adding the alias: split() is the obvious =
name for this function, and as you say it would mirror join()

Cheers,

Rowan Tommins
[IMSoP]