# Class names (with regard to autoload) - Thread: https://externals.io/message/43191 - Original archive: https://news-web.php.net/php.internals/43191 - Started by Robin Burchell on 2009-02-26T20:27:36+00:00 - Messages: 6 - Reply by email to internals@lists.php.net - More for agents: search with https://externals.io/search.md?q={query} · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt --- ## #43191 — Robin Burchell — 2009-02-26T20:27:36+00:00 URL: https://externals.io/message/43191#43191 Hi all, I've been playing around with autoload lately, and specifically using autoload to load classes which may be in subdirectories. I came up with the following test script: The first line demonstrates what I'm after ("autoloading Foo.Bar"), and the second seems to give a bit of an odd/incorrect (in these circumstances) result(__autoload is invoked with "Foo"). Is it possible to reconcile this behaviour to be a bit more consistant, and provide the behaviour I'm after? I'm willing to try provide a patch, if someone would be kind enough to direct me to the part(s) of the engine I'd need to look at. Thanks, Robin --- ## #43192 — Guilherme Blanco — 2009-02-26T20:48:40+00:00 URL: https://externals.io/message/43191#43192 · In reply to: #43191 I doubt you'll find a patch to it. Mainly, the patch will be against this principle: Hi all, > > I've been playing around with autoload lately, and specifically using > autoload to load classes which may be in subdirectories. I came up > with the following test script: > > > function __autoload($sName) > { >        echo "autoloading " . $sName . "\n"; > } > > $s = "Foo.Bar"; > > $a = new $s(); > $b = new Foo.Bar(); > ?> > > The first line demonstrates what I'm after ("autoloading Foo.Bar"), > and the second seems to give a bit of an odd/incorrect (in these > circumstances) result(__autoload is invoked with "Foo"). > > Is it possible to reconcile this behaviour to be a bit more > consistant, and provide the behaviour I'm after? > > I'm willing to try provide a patch, if someone would be kind enough to > direct me to the part(s) of the engine I'd need to look at. > > Thanks, > Robin > > -- > > > > -- Guilherme Blanco - Web Developer CBC - Certified Bindows Consultant Cell Phone: +55 (16) 9215-8480 MSN: guilhermeblanco@hotmail.com URL: http://blog.bisna.com São Paulo - SP/Brazil --- ## #43193 — Stuart — 2009-02-26T21:43:43+00:00 URL: https://externals.io/message/43191#43193 · In reply to: #43191 2009/2/26 Robin Burchell > Hi all, > > I've been playing around with autoload lately, and specifically using > autoload to load classes which may be in subdirectories. I came up > with the following test script: > > > function __autoload($sName) > { > echo "autoloading " . $sName . "\n"; > } > > $s = "Foo.Bar"; > > $a = new $s(); > $b = new Foo.Bar(); > ?> > > The first line demonstrates what I'm after ("autoloading Foo.Bar"), > and the second seems to give a bit of an odd/incorrect (in these > circumstances) result(__autoload is invoked with "Foo"). > > Is it possible to reconcile this behaviour to be a bit more > consistant, and provide the behaviour I'm after? > > I'm willing to try provide a patch, if someone would be kind enough to > direct me to the part(s) of the engine I'd need to look at. > The concatenation operator (.) is not valid in a class name. Use an underscore and you'll be able to get the behaviour you want. -Stuart -- http://stut.net/ --- ## #43194 — Robin Burchell — 2009-02-26T21:51:39+00:00 URL: https://externals.io/message/43191#43194 · In reply to: #43192 On Thu, Feb 26, 2009 at 8:48 PM, Guilherme Blanco wrote: > I doubt you'll find a patch to it. > > Mainly, the patch will be against this principle: > Hmm, that wouldn't actually have an impact, as (at least, what I'm aiming for/wanting to do) is not to introduce periods as valid in actual class names, but to allow them to be used for autoloading - so I can have something like this (abbreviated terribly, apologies): function __autoload($sFoo) { include(str_replace(".", "/", $sFoo)); } .. so, your example would expand to Foo.Bar, which given that periods aren't allowed in a class name doesn't exist as a class, and won't work. Unless I'm missing something? Actually, I don't believe that expanding is even done - i.e. it doesn't respect defines there Seems that way on testing, too: class FooBar { // ... } define('Part', 'Foo'); define('SubPart', 'Bar'); $fb = new Part.SubPart(); gives: Fatal error: Class 'Part' not found .. but that part of things is a going off on a bit of a tangent. --- ## #43195 — Lewis Wright — 2009-02-26T22:04:40+00:00 URL: https://externals.io/message/43191#43195 · In reply to: #43194 I think what you're doing is effectively the same as doing: $fb = new Part .''. SubPart(); So basically concatenating a new instance of Part, with the results from the function SubPart. Try defining Part as a class and see what error you get. If I'm right, you'll get an undefined function error. 2009/2/26 Robin Burchell : > On Thu, Feb 26, 2009 at 8:48 PM, Guilherme Blanco > wrote: > > […] > > > Hmm, that wouldn't actually have an impact, as (at least, what I'm > aiming for/wanting to do) is not to introduce periods as valid in > actual class names, but to allow them to be used for autoloading - so > I can have something like this (abbreviated terribly, apologies): > > function __autoload($sFoo) > { >    include(str_replace(".", "/", $sFoo)); > } > > .. so, your example would expand to Foo.Bar, which given that periods > aren't allowed in a class name doesn't exist as a class, and won't > work. Unless I'm missing something? > > > > Actually, I don't believe that expanding is even done - i.e. it > doesn't respect defines there > > Seems that way on testing, too: > > class FooBar > { >        // ... > } > > define('Part',  'Foo'); > define('SubPart',  'Bar'); > > $fb = new Part.SubPart(); > > > gives: Fatal error: Class 'Part' not found > > .. but that part of things is a going off on a bit of a tangent. --- ## #43211 — karpeles@ookoo.org — 2009-02-27T06:08:11+00:00 URL: https://externals.io/message/43191#43211 · In reply to: #43191 Le jeudi 26 février 2009 à 20:27 +0000, Robin Burchell a écrit : > I've been playing around with autoload lately, and specifically using > autoload to load classes which may be in subdirectories. I came up > with the following test script: The best solution for you is to wait for PHP 5.3.0, then use: function __autoload($class) { $path = CLASSES_DIR . '/' . str_replace($class, '\\', '/', $class) . '.class.php'; include($path); } Namespaces are probably the best way to achieve this (apart from using _). Mark