Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:100861
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain pmmaga.net designates 149.210.149.73 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <CAF+90c_hcR7VV1nvpH8f3e+H9oarUDG0B=ixvtf+ZkT9QL3UaA@mail.gmail.com>
References: <9a3f15dd-5dfc-144b-cc5f-66f689c33f53@php.net> <CAF+90c_hcR7VV1nvpH8f3e+H9oarUDG0B=ixvtf+ZkT9QL3UaA@mail.gmail.com>
Date: Wed, 11 Oct 2017 10:03:21 +0100
Message-ID: <CAJd4=YaUjs+NaDvoNcS0kFu9_FuuVaSs-9YYxGB8v1OuxJzQig@mail.gmail.com>
To: Nikita Popov <nikita.ppv@gmail.com>
Cc: PHP internals <internals@lists.php.net>
Content-Type: multipart/alternative; boundary="001a1140f96c47749d055b41b372"
Subject: Re: [PHP-DEV] Strict type declarations not enforced for Reflection
 API invocation
From: mail@pmmaga.net (=?UTF-8?Q?Pedro_Magalh=C3=A3es?=)

--001a1140f96c47749d055b41b372
Content-Type: text/plain; charset="UTF-8"

On Tue, Oct 10, 2017 at 5:53 PM, Nikita Popov <nikita.ppv@gmail.com> wrote:
>
> The problem are not internal function calls, the problem are callbacks. In
> fact, the proposed fix does not actually fix the problem you encountered in
> PHPUnit, as it is going to use the strictness mode at the reflection
> call-site, not the strictness mode used by the file defining the data
> provider.
>

For those that didn't have a look at the PR, my goal was to try to ensure
that if we are looking for which strictness to use, we should look at the
place of the closest "user" call instead of the function (user or internal)
that called the current one. The reason for this problem lies in the fact
that `ZEND_ARG_USES_STRICT_TYPES()` will simply look at the caller
regardless of what it is. Although this is enough to allow calling
functions that were defined as non-strict in a strict manner, it does cause
this kind of shortcomings with callbacks.


> I believe that the proper way to fix this is to handle dynamic function
> invocations differently from direct invocations. Direct invocations should
> use the strictness level of the call-site, while dynamic invocations should
> use the strictness level of the declaration-site.


I agree that this should be about dynamic vs direct instead of user vs
internal but IMHO, making the strictness vary from call-site to
declaration-site depending on that may be a bit too confusing. Would it be
possible/reasonable to try to find the place where the dynamic call was
started? (i.e. the dataProvider, the mapped function for a userland
array_map, etc...)

Regards,
Pedro

--001a1140f96c47749d055b41b372--