Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:105765
Content-Type: text/plain;
	charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.8\))
In-Reply-To: <DM5PR06MB2857A407C06D8243AF429DC4DE060@DM5PR06MB2857.namprd06.prod.outlook.com>
Date: Mon, 20 May 2019 21:36:33 +0200
Cc: PHP internals <internals@lists.php.net>
Content-Transfer-Encoding: quoted-printable
Message-ID: <C6EBFEA1-779C-47FA-8C13-838F58A5EE95@gmail.com>
References: <DM5PR06MB285728875FF1272F25FD35E7DE060@DM5PR06MB2857.namprd06.prod.outlook.com>
 <16ad66607a0.27c1.08be835b7d1a2c2edb4c4286afe1a236@gmail.com>
 <DM5PR06MB2857A407C06D8243AF429DC4DE060@DM5PR06MB2857.namprd06.prod.outlook.com>
To: Theodore Brown <theodorejb@outlook.com>,
 Marco Pivetta <ocramius@gmail.com>
Subject: Re: [PHP-DEV] High performance function autoloading
From: gadelat@gmail.com (Gabriel O)

On 20. May 2019, at 20:42, Theodore Brown <theodorejb@outlook.com> =
wrote:
>=20
> Can you clarify? I'm not sure what performance improvement trick you
> are referring to, or what you mean by "importing all functions". Are
> you saying it's a misconception that triggering an autoloader on
> unqualified function calls in a namespace will slow down performance?
> As I understand it, this is the roadblock for function autoloading.
>=20
> Best regards,
> Theodore

Yes, this is misconception. Namespace resolutions are cached. First =
fixer tool bringing support for this was correct: PHP-CS-Fixer auto =
imports only optimized functions for this very same reason.


> On 20. May 2019, at 20:45, Marco Pivetta <ocramius@gmail.com> wrote:
>=20
> Disabling function mocking is good =C2=A9=EF=B8=8F
>=20
> It was a terrible practice in first place, and it is usually done for
> impure functions that should be wrapped in integration-tested =
adapters.


=E2=80=A6which dramatically affects design of code. Nobody builds =
injectable replacements for these calls. As always, you are speaking =
from POV of library maintainer. As a user of your library, you don=E2=80=99=
t have perfect design. Nobody has.=20

As a user of the library, I want to have ability to replace every aspect =
of it without copying it, while preserving simplicity - this is =
impossible to do, because these goals are in opposition.=20

As a contributor of the library, I have found it impossible to write =
tests for some parts of very same library, because its code style guide =
dictates importing all the functions which makes it impossible to mock =
function call. And replacing them with injectable adapters was deemed =
overkill (and is subject of BC concerns).

Anyways, it doesn=E2=80=99t matter. Disabling function fallback for all =
functions has very little benefit like I mentioned.=