Bug #68792 [Dup->Opn]: Can't declare closure using $this within static method (context)

Edit report at https://bugs.php.net/bug.php?id=68792&edit=1 ID: 68792 Updated by: nikic@php.net Reported by: llmll at gmx dot de Summary: Can't declare closure using $this within static method (context) -Status: Duplicate +Status: Open Type: Bug Package: Scripting Engine problem Operating System: any PHP Version: 5.5.20 -Assigned To: levim +Assigned To: nikic Block user comment: N Private report: N Previous Comments: ------------------------------------------------------------------------ [2015-01-13 07:46:51] llmll at gmx dot de Don't take it personal. I will try to present the facts for why your patch fixes the right thing the wrong way. static:: implements late static binding, so it always returns the class name of where it is referenced - regardless of the calling context, whether it is from an instance A$->foo() or a class method A::foo(). class A { public function foo() { echo static::class; } } class B extends A {} B::foo(); //prints "B" (new B)->foo(); // prints "B" Your patch obviously fixed the late static binding problem for closures. But it introduced a wrong behaviour by misinterpreting the static keyword from the containing method. It should work without inspecting it. ------------------------------------------------------------------------ [2015-01-13 02:52:44] levim@php.net > I really mean no offense, but this fix of yours is against all practical logic. This is still being hostile; please stop. As for static closures: they already existed. A static closure in PHP means that you don't capture the $this context. Again, I did not add this feature. > What could ever be the reason to make a closure depend on the declaring context? Look at this case (http://3v4l.org/105nh): <?php class A { static function foo() { $f = function() { return static::class; }; return $f(); } } class B extends A {} var_dump(B::foo()); ?> Before I patched it the result was string(1) "A" instead of string(1) "B". Is this any less valid that the code you want to work? I'm not suggesting they can't both be supported, but as it stands it how things are. ------------------------------------------------------------------------ [2015-01-12 21:57:38] llmll at gmx dot de I really mean no offense, but this fix of yours is against all practical logic. Someone wanted static closures? Well, closures are static by definition. Atomic, independent pieces of code which can run in any context, thats the whole point of having closures. Convieniently, PHP allows to bind a closure to some object instance during runtime, which simplifies the use of $this. static:: then should point to the executing class context if available, we know that as late static binding. What could ever be the reaseon to make a closure depend on the declaring context? That really gives me gray hair. I humbly think this is absolutely wrong and should be re-fixed. However, thank you for reading and spending thoughts on this matter. ------------------------------------------------------------------------ [2015-01-12 21:31:14] levim@php.net > I really begin to wonder if you ever accept something is wrong, bogus, inconsistent or simply > stupid. You are being hostile. I am the person who made closures in static methods implicitly static as well, and I did so because of a bug report. So yes, I do accept things as bugs. Please be less accusatory in the future. ------------------------------------------------------------------------ [2015-01-12 16:00:26] llmll at gmx dot de You have to try the second example from my comment. There is a working snippet. Re-enable the commented "static" and it fails. That must not happen, as the static only tags the enclosing method, not the closure itself. I really begin to wonder if you ever accept something is wrong, bogus, inconsistent or simply stupid. Claiming that a clear bug happens the same way as last year doesn't make it any better for anybody. ------------------------------------------------------------------------ The remainder of the comments for this report are too long. To view the rest of the comments, please view the bug report online at https://bugs.php.net/bug.php?id=68792 -- Edit this bug report at https://bugs.php.net/bug.php?id=68792&edit=1