Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:100419
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.223.193 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <BY2PR02MB2989D1A6035072A57D098DDAC970@BY2PR02MB298.namprd02.prod.outlook.com>
References: <BY2PR02MB2989D1A6035072A57D098DDAC970@BY2PR02MB298.namprd02.prod.outlook.com>
Date: Wed, 6 Sep 2017 15:19:07 +0200
Message-ID: <CAF+90c-mHpF8N7tUiggUkO0m2Pxevz3OZ70TBKc1gFkZ2abzPQ@mail.gmail.com>
To: Zeev Suraski <zeev@zend.com>
Cc: Arvids Godjuks <arvids.godjuks@gmail.com>, Dan Ackroyd <danack@basereality.com>, 
	"internals@lists.php.net" <internals@lists.php.net>
Content-Type: multipart/alternative; boundary="94eb2c18a11851092005588530bd"
Subject: Re: [PHP-DEV] Providing built-in functionality written in PHP (was
 RE: [PHP-DEV] [VOTE] UUID)
From: nikita.ppv@gmail.com (Nikita Popov)

--94eb2c18a11851092005588530bd
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Wed, Sep 6, 2017 at 2:46 PM, Zeev Suraski <zeev@zend.com> wrote:

> I think that actually makes a lot of sense, and not just because of the
> supportability =E2=80=93 but also because of security.  A whole class of =
security
> exploits =E2=80=93 buffer/stack overflows, underruns and all sorts of mem=
ory
> mismanagement become irrelevant when the code is implemented in PHP.  I
> brought this direction up in a discussion on the Security mailing list a
> few weeks ago without any traction =E2=80=93 but it probably makes more s=
ense to
> discuss it here anyway.
>
> I think that currently, there are two main challenges:
>
>   1.  Performance =E2=80=93 compute intensive logic is way slower in PHP =
compared
> to C.
>   2.  Delivery method  =E2=80=93  we don=E2=80=99t currently have a good =
way of providing
> functions that are written in PHP and have them provide the same =E2=80=
=98native=E2=80=99 /
> =E2=80=98builtin=E2=80=99 experience as functions/classes written in C.
>
> Regarding #1, often this isn=E2=80=99t very important as not all pieces o=
f code
> are that compute intensive.  Moreover, if/when JIT materializes, compute
> intensive logic in PHP will become a lot faster than it is today and
> probably in the same ballpark as C =E2=80=93 so it=E2=80=99ll open the do=
or for us
> implementing more and more things in PHP.
>
> Regarding #2 =E2=80=93 I think that=E2=80=99s something that can be solve=
d relatively
> easily, but admittedly I haven=E2=80=99t completely thought it through (r=
ead: I
> barely thought about it).
>
> We could create a mechanism where the contents of certain .php files is
> embedded into the binary, compiled during MINIT, and made available prett=
y
> at the same =E2=80=98builtinness=E2=80=99 level as C extensions.   We=E2=
=80=99d probably have to be
> pretty selective in terms of what goes in there =E2=80=93 probably just a=
s
> selective as we are with the C-based extensions, but I=E2=80=99d imagine =
that
> things like ext/exif, UUID, and perhaps even things like unserialize()
> could find themselves written in pure PHP using such a mechanism.
>
> Thoughts?
>
> Zeev
>

There has been a discussion about this recently:
https://externals.io/message/99366

Nikita

--94eb2c18a11851092005588530bd--