Am 10.10.2017 um 18:41 schrieb Niklas Keller:
2017-10-10 17:10 GMT+02:00 li...@rhsoft.net <mailto:li...@rhsoft.net>
<li...@rhsoft.net <mailto:li...@rhsoft.net>>:
Am 10.10.2017 um 16:57 schrieb Christoph M. Becker:
On 10.10.2017 at 15:41, Sebastian Bergmann wrote:
I consider this a serious bug that leads to unexpected,
confusing problems
such as
https://github.com/sebastianbergmann/phpunit/issues/2796#issuecomment-335180273
<https://github.com/sebastianbergmann/phpunit/issues/2796#issuecomment-335180273>.
I understand Nikita's point of view (see
https://github.com/php/php-src/pull/2837#issuecomment-335405067
<https://github.com/php/php-src/pull/2837#issuecomment-335405067>)
that
changing this behavior (aka. fixing this bug) can be
considered a
"non-trivial backwards compatibility break". Therefore I
would like to
bring this issue to the attention of this list with this mail.
This is most certainly not an *implementation* bug. The RFC which
introduced strict typing[1] states:
| By default, all PHP files are in weak type-checking mode.
yes but the file in question has strict-types enabled
declare(strict_types=1);
Yes, but `array_map` also uses weak types for the callback, like any
other internal function call.
But `call_user_func` is also special-cased, maybe we should do the same
with the reflection calls.
All in all, two typing modes were a bad idea to begin with, mostly
because nobody payed attention to callbacks
the two typing modes are perfect and when you converted a 15 years old
codebase to strcit-ytpes and type-hints everywehere you know why - first
you start with typehints and you can't do that strict or you will burn
out and give up
but when i define strict types in a PHP file everything but code in
includes has to be strict typed inherited
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
