Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:112592
MIME-Version: 1.0
References: <CABTDVhA-FvEXdS+A=T5bJnX5YxjseN+ko14GUOqnuA52-omiiw@mail.gmail.com>
 <CABTDVhCKu9hjspzs3vv7=9=4C=NhSD9+85e++RhY6wXUwXGc6A@mail.gmail.com>
 <1f319ee0-e5e2-3d5e-2b8c-4d3c073f98e6@gmx.de> <CABTDVhCY44bT=U489KueCV3309AuN+1zPjRcfSSMEvfcM9Yfkw@mail.gmail.com>
 <CAGgaK7+YXrZg5LR6a_BFS=BGEPNj023JjS+wQpymas4H9zGBBQ@mail.gmail.com>
In-Reply-To: <CAGgaK7+YXrZg5LR6a_BFS=BGEPNj023JjS+wQpymas4H9zGBBQ@mail.gmail.com>
Date: Wed, 23 Dec 2020 00:02:09 +0900
Message-ID: <CABTDVhB2j--mS8DzQy6VNYu-DNZRuYK5Ex-VDOucbt4=rk8ZNA@mail.gmail.com>
To: Max Semenik <maxsem.wiki@gmail.com>, internals@lists.php.net
Content-Type: multipart/alternative; boundary="0000000000009e301105b70edcbd"
Subject: Re: [PHP-DEV] Re: Improving PRNG implementation.
From: zeriyoshi@gmail.com (zeriyoshi)

--0000000000009e301105b70edcbd
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Thanks. I have edited the RFC based on your question.

> Method names: you don't need to emulate the conventions from non-OOP
code, e.g. string_shuffle() --> strShuffle and array_rand() -->
arrayRand(). Use a more naturally readable convention like shuffleString().

As mentioned in the RFC, it is intended for interoperability. However, you
may be right that we should forget the past and aim for a better
implementation.

> Do you have evidence of widespread demand for this functionality? Usage
statistics for the userland implementation you mentioned[3] don't look that
hot.

A difficult problem. At least for PHP, which is a web-oriented language,
there may be little demand for it.

In my opinion, an object-scoped implementation will be needed in the
future. For example, there are currently proposed implementations of Fiber,
which will certainly make global state unpredictable. Swoole extension is
also a concern, although I am not familiar with this.

Regards,
Go Kudo

2020=E5=B9=B412=E6=9C=8820=E6=97=A5(=E6=97=A5) 23:09 Max Semenik <maxsem.wi=
ki@gmail.com>:

> On Sun, Dec 20, 2020 at 6:45 AM zeriyoshi <zeriyoshi@gmail.com> wrote:
>
>> Thanks cmb. I have created a first draft of an RFC. I think I've covered
>> all the necessary requirements, but I'd like to know if there are any
>> problems.
>> https://wiki.php.net/rfc/object_scope_prng
>
>
> A few IMHO comments:
> * Namespaces are controversial. There were multiple discussions and RFCs
> about introducing them for new functionality but nothing came out of it.
> Still probably worth adding an option for \PHP\PRNG to the vote, or
> something like that.
> * The "Which set of PRNGs should be provided as standard?" vote seems
> vague to me. Are you proposing to add only one algorithm? Why? Also, the
> adding just interfaces option makes no sense because there's Composer for
> things that don't require a C implementation.
> * "One possible solution is to implement the PRNG in pure PHP. There is
> actually a userland library [1], but it is not fast enough for PHP at the
> moment. (However, this may be improved by JIT)" - this definitely needs a
> comparison because implementation in C is only needed for performance
> reasons.
> * Method names: you don't need to emulate the conventions from non-OOP
> code, e.g. string_shuffle() --> strShuffle and array_rand() -->
> arrayRand(). Use a more naturally readable convention like shuffleString(=
).
> * The return value in "function shuffle(array &$array): bool" means that
> if something goes wrong, callers will have no idea what it is. Just throw
> an exception.
> * If the classes are namespaced, don't prefix their names,
> e.g. PRNGInterface --> RandomGenerator (yay, so many opportunities for
> bikeshedding!)
> * Perhaps it's worth mentioning other languages using this pattern, e.g.
> Java[1] and C#[2].
> * Do you have evidence of widespread demand for this functionality? Usage
> statistics for the userland implementation you mentioned[3] don't look th=
at
> hot.
>
> ------
> [1] https://docs.oracle.com/javase/8/docs/api/java/util/Random.html
> [2] https://docs.microsoft.com/en-us/dotnet/api/system.random?view=3Dnet-=
5.0
> [3] https://packagist.org/packages/savvot/random
>
>

--0000000000009e301105b70edcbd--