>Пятница, 31 декабря 2021, 10:20 +03:00 от Christian Schneider
>:
>
>Am 31.12.2021 um 03:21 schrieb Kirill Nesmeyanov < n...@xakep.ru >:
>> I support this behavior fix because in its current form, due to a similar
>> problem (almost?), all PSR-7 implementations contain bugs that violate
>> RFC7230 (section 3.2:
>> https://datatracker.ietf.org/doc/html/rfc7230#section-3.2 ). Thus,
>> physically, by the standard, all headers can have the name "0" (like «0:
>> value»), but when stored inside implementations, it is converted to a string
>> and a problem arises ($message->getHeaders() // returns array> string> instead of array).
>>
>> To solve this problem without BC, it seems to me that Nikita's idea with
>> «declares» is most suitable, like:
>> ```
>> > declare(strict_types=1, strict_array_keys=1);
>>
>> $array["0"] = 42; // array(1) { string(1) "0" => int(42) }
>> ```
>
>I probably sound like a broken record but I'd strongly advise against it as it
>creates language dialects which means a line like
>$a["0"] = 42;
>behaves differently depending on context, e.g. copying code from Stackoverflow
>might lead to subtle bugs and reminds me way too much of the ini-settings
>magic_quotes_gpc.
>
>My take is that changing behavior is ok if it can go through a
>deprecation/warning phase but that is IMHO not the case here.
>
>If people want to have different behavior I'd rather have a new type like
>$a = new Dict;
>$a["0"] = 42;
>possibly with a short-hand for initialization.
>
>Regards,
>- Chris
>
>--
>PHP Internals - PHP Runtime Development Mailing List
>To unsubscribe, visit: https://www.php.net/unsub.php
Yep =(
In any case, I don't see any other solutions to the problem yet... Well,
perhaps adding the Map and Set objects like in ur example.
On the other hand, probably we can forget about the BC and still fix it in
major release. I don't think that anyone really uses such functionality and
this is part of the real business logic. This is what I personally would
prefer. Then this will automatically fix all PSR-7 implementations, and I hope
there is a lot of code where such a case with automatic type conversion is not
taken into account.
--
Kirill Nesmeyanov
