Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:109317
Content-Type: text/plain;
	charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 13.0 \(3608.60.0.2.5\))
In-Reply-To: <CAAKU0ukbJyCt-N_Q5uCY0no4QOe37TKUUMvbjDwNfTM+3VRczw@mail.gmail.com>
Date: Thu, 26 Mar 2020 02:52:27 +0100
Cc: PHP internals <internals@lists.php.net>
Content-Transfer-Encoding: quoted-printable
Message-ID: <B6D7B5D8-F7B5-403A-BC60-6F48A2BEA6EF@gmail.com>
References: <CAAKU0ukbJyCt-N_Q5uCY0no4QOe37TKUUMvbjDwNfTM+3VRczw@mail.gmail.com>
To: Levi Morrison <levi.morrison@datadoghq.com>
Subject: Re: [PHP-DEV] Removing warning for failed include
From: claude.pache@gmail.com (Claude Pache)



> Le 26 mars 2020 =C3=A0 01:11, Levi Morrison via internals =
<internals@lists.php.net> a =C3=A9crit :
>=20
> It's bothered me for quite some time that a failed include emits a
> warning. This is because it's by design that the author chose
> `include` and not `require`. It's _expected_ that it may fail and they
> are okay with that.
>=20
> As an example, consider a classic autoloading case. It could be as
> simple and performant as:
>=20
>    $file =3D /* something derived from class name */;
>    include $file;
>=20
> But because of the warning, in practice authors tend to use
> `file_exists` + require instead:
>=20
>    $file =3D /* something derived from class name */;
>    if (file_exists($file)) {
>        require $file;
>    }
>=20
> Weird isn't it? Authors are using require instead of include for the
> case where failure is tolerated. This is a clear signal to me that
> include isn't doing its job. The warning gets in the way.
>=20
> Any reasons we shouldn't just remove this warning for PHP 8?
>=20

A quick look on the code base I=E2=80=99m taking care of, reveals that =
most `include` instructions expect that the file will be in fact =
included. If it isn=E2=80=99t, the error handler will be triggered and =
send me a bug report. (Yes I know, all those `include` could have been =
spelled `require`.)

On the other hand, in the occasional cases where I do want to include a =
file only if it exists, I wouldn=E2=80=99t lightheartedly remove the =
`file_exists()` check, because its presence makes the intent of the code =
clearer.

=E2=80=94Claude=