Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:105766
To: Gabriel O <gadelat@gmail.com>, Marco Pivetta <ocramius@gmail.com>
CC: PHP internals <internals@lists.php.net>
Thread-Topic: [PHP-DEV] High performance function autoloading
Thread-Index: AQHVDy+ySrd8q1/GuEiFhoiHIw5ZMaZ0TZWAgAABNxaAABl9gIAAA/j2
Date: Mon, 20 May 2019 20:30:21 +0000
Message-ID: <DM5PR06MB285779ADEA21ED467FBDBEA0DE060@DM5PR06MB2857.namprd06.prod.outlook.com>
References: <DM5PR06MB285728875FF1272F25FD35E7DE060@DM5PR06MB2857.namprd06.prod.outlook.com>
 <16ad66607a0.27c1.08be835b7d1a2c2edb4c4286afe1a236@gmail.com>
 <DM5PR06MB2857A407C06D8243AF429DC4DE060@DM5PR06MB2857.namprd06.prod.outlook.com>,<C6EBFEA1-779C-47FA-8C13-838F58A5EE95@gmail.com>
In-Reply-To: <C6EBFEA1-779C-47FA-8C13-838F58A5EE95@gmail.com>
Accept-Language: en-US
Content-Language: en-US
x-ms-exchange-messagesentrepresentingtype: 1
x-tmn: [JI7IsVuuxWdYXT5AtsLKbGvfiyl+iXAY]
x-ms-publictraffictype: Email
x-incomingheadercount: 44
x-eopattributedmessage: 0
x-microsoft-antispam: BCL:0;PCL:0;RULEID:(2390118)(5050001)(7020095)(20181119110)(201702061078)(5061506573)(5061507331)(1603103135)(2017031320274)(2017031323274)(2017031324274)(2017031322404)(1601125500)(1603101475)(1701031045);SRVR:SN1NAM04HT153;
x-ms-traffictypediagnostic: SN1NAM04HT153:
x-ms-exchange-purlcount: 2
x-microsoft-antispam-message-info: Mh242OMOKPPHbc+BzIwEilZ8mQ+kqg+/y6ctvo1lOZCpJK8mhPYrGRamqq3nAcW6JHWTDF0TfQH5IAYTb6288jduecC/WuMVPjzFdztQ75AHXjKSsxe6Hxi8GwcfIboPKMoCHyGzDtdD6lIRBy3xc4cxM2YbQgxyQRZJnc1/DL+Z9jLMCQmiDj+lpGLeVHZ7
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-OriginatorOrg: outlook.com
X-MS-Exchange-CrossTenant-RMS-PersistedConsumerOrg: 00000000-0000-0000-0000-000000000000
X-MS-Exchange-CrossTenant-Network-Message-Id: aefe1910-872d-4780-59a8-08d6dd61fafa
X-MS-Exchange-CrossTenant-rms-persistedconsumerorg: 00000000-0000-0000-0000-000000000000
X-MS-Exchange-CrossTenant-originalarrivaltime: 20 May 2019 20:30:21.1260
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Internet
X-MS-Exchange-CrossTenant-id: 84df9e7f-e9f6-40af-b435-aaaaaaaaaaaa
X-MS-Exchange-Transport-CrossTenantHeadersStamped: SN1NAM04HT153
Subject: Re: [PHP-DEV] High performance function autoloading
From: theodorejb@outlook.com (Theodore Brown)

On Mon, May 20, 2019 at 2:36 PM Gabriel O <gadelat@gmail.com> wrote:=0A=
=0A=
> > Can you clarify? I'm not sure what performance improvement trick you=0A=
> > are referring to, or what you mean by "importing all functions". Are=0A=
> > you saying it's a misconception that triggering an autoloader on=0A=
> > unqualified function calls in a namespace will slow down performance?=
=0A=
> > As I understand it, this is the roadblock for function autoloading.=0A=
> =0A=
> Yes, this is misconception. Namespace resolutions are cached. First=0A=
> fixer tool bringing support for this was correct: PHP-CS-Fixer auto=0A=
> imports only optimized functions for this very same reason.=0A=
=0A=
I'm confused, since this doesn't match what I've read elsewhere. For=0A=
example, Nikita wrote the following last time function autoloading was=0A=
discussed: [1]=0A=
=0A=
> Calling the autoloader if a global function with the name exists will=0A=
> totally kill performance. This means that every call to strpos() or=0A=
> any of the other functions in the PHP standard library will have to=0A=
> go through the autoloader first, unless people use fully qualified=0A=
> names (which, currently, they don't). This is completely out of the=0A=
> question.=0A=
=0A=
In the same thread, it was suggested that a cache could be used=0A=
to avoid repeatedly triggering the autoloader, but concerns were=0A=
also pointed out with this: [2]=0A=
=0A=
> > Of course calling e.g. strpos() should not trigger the auto-loader=0A=
> > repeatedly - can we cache the information that the auto-loader was=0A=
> > attempted once during the current script execution? so that e.g.=0A=
> > only the first call to strpos() triggers the auto-loader?=0A=
> =0A=
> I think triggering it even once for every internal function in the=0A=
> code may be too much.=0A=
> =0A=
> > I suppose it would still happen once for every namespace from which=0A=
> > strpos() gets called, so maybe this optimization doesn't help much.=0A=
> =0A=
> Exactly. Also, caching stuff assumes static environment. What if it=0A=
> changes, e.g. autoloader gets different configuration?=0A=
=0A=
This is why I suggested the alternative of by default falling back=0A=
to the global namespace for unqualified functions before triggering=0A=
the autoloader, and allowing this fallback to be disabled for the rare=0A=
case where functions in a namespace are spread across files and need=0A=
to override a global function.=0A=
=0A=
I don't think this proposal changes anything related to function=0A=
mocking. As you pointed out, libraries can already prevent mocking=0A=
by importing or qualifying global function calls.=0A=
=0A=
Sincerely,=0A=
Theodore=0A=
=0A=
[1]: https://externals.io/message/94895#94897=0A=
[2]: https://externals.io/message/94895#94956=