On Thu, Nov 20, 2014 at 2:43 AM, Johannes Schlüter
<[email protected]> wrote:
> On Wed, 2014-11-19 at 14:33 +0000, Alain Williams wrote:
>> How many servers are stuck on PHP 4 ?
>>
>> Of those 'stuck' servers, how many have applications still under active
>> development ?
>>
>> The point is: how many people would get annoyed if PEAR stopped supporting PHP 4 ?
>
> The point about breaking this is *not* PHP 4 compatibility but
> compatibility between PECL packages. The name of the constructor method
> is part of the API.
>
> Think about code like this in module A:
>
> class A_class {
> function A_class() { }
> }
>
> and then module B extending this:
>
> class B_class extends A_class {
> function B_class() {
> A_class();
> }
> }
>
> I also wonder how Andrea's tool handles more indirect cases (C_class
> extends B_class, while B_class has no ctor so C_class calls directly
> A_class's ctor)
>
> So I'm -1 on this.
I just want to make sure I understand you correctly: you are saying
you are voting no on this RFC because a tool, which is not part of
this RFC but we kindly provide, doesn't detect when a certain thing is
called?