Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:114528
To: PHP internals <internals@lists.php.net>
References: <CABTDVhAd8r_OPsdbzxrFeZTZEHuZjKyPfcYhNcdfdSZWbUJ-Lw@mail.gmail.com>
 <72767cd7-ffc5-c84b-523c-123ff6ef30fa@gmail.com>
 <CABTDVhBD_D-UF-c6=5sce=p8QsOaFZ36jK1cgz9Gy0SW3JrF2g@mail.gmail.com>
Message-ID: <d1611452-e061-bbac-3ebe-e1d65e521546@gmail.com>
Date: Thu, 20 May 2021 09:45:41 +0100
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Thunderbird/78.10.1
MIME-Version: 1.0
In-Reply-To: <CABTDVhBD_D-UF-c6=5sce=p8QsOaFZ36jK1cgz9Gy0SW3JrF2g@mail.gmail.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Content-Language: en-GB
Subject: Re: [PHP-DEV] [RFC] [Draft] Add RNG extension and deprecate
 mt_srand()
From: rowan.collins@gmail.com (Rowan Tommins)

On 20/05/2021 05:59, Go Kudo wrote:
> Thanks.
>
> Deprecation will be done only to prevent unintended MT state dependent 
> code. We don't plan to go as far as removing the implementation for now.


I think every deprecation notice should mean a plan to remove. If you 
just want to encourage people to discover the new functions, I'm not 
sure a message in their logs every time mt_srand runs is the right way 
to do it.


> > Firstly, making these functions independent of mt_srand() is a 
> breaking change, so cannot happen until PHP 9.0 at the earliest.
>
> To be honest, I don't understand how far PHP is willing to go to 
> accept disruptive changes.


The official policy is here: https://wiki.php.net/rfc/releaseprocess It 
relies on the phrase "backwards compatibility", which is admittedly 
quite hard to define.

You are right that the change from using rand() to using mt_rand() in 
7.1 is arguably not backwards compatible. However, the following code 
can be used to reproducibly shuffle two arrays in any version of PHP:

srand(123456);
shuffle($array1);
shuffle($array2);

The exact sequence returned will be different in 7.1 from 7.0, but 
because srand() is now an alias of mt_srand(), it still affects the 
shuffle() function in the expected way. If we switch shuffle() to an 
algorithm that can't be seeded at all, this code will simply stop 
working, and have to be completely rewritten.

A "polyfill" of sorts is possible, by storing an instance of the new 
RNG\Randomizer in a global or static variable, but it's more than just a 
find-and-replace, especially since many users will need to support 
version 8.0 and 8.1 in the same code base.

As I said previously, I'm not convinced these functions need to stop 
using the MT algorithm at all, and certainly see no reason to rush the 
change.


Regards,

-- 
Rowan Tommins
[IMSoP]