Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:85441
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.192.172 as permitted sender)
Content-Type: multipart/alternative;
	boundary=Apple-Mail-CC8DB9E7-BA19-4E33-BAFB-705AB1FCB7FF
Mime-Version: 1.0 (1.0)
In-Reply-To: <CAF+90c9Kw0iGxEZkpwbfNV+uZKrnpJ32gz9g257w00G1-rC1JQ@mail.gmail.com>
Date: Tue, 24 Mar 2015 17:28:48 +0800
Cc: Dmitry Stogov <dmitry@zend.com>, Xinchen Hui <xinchen.h@zend.com>,
 Joe Watkins <pthreads@pthreads.org>,
 PHP Internals <internals@lists.php.net>
Content-Transfer-Encoding: 7bit
Message-ID: <ADC66A6B-039B-4AEA-B6D9-92AF2DEE46A7@gmail.com>
References: <CA+9eiLsAQPTkVuR=D0wXMFXhBNM8DQzGq0DRXzSuL+hjPsUpgA@mail.gmail.com> <CAF+90c9Kw0iGxEZkpwbfNV+uZKrnpJ32gz9g257w00G1-rC1JQ@mail.gmail.com>
To: Nikita Popov <nikita.ppv@gmail.com>
Subject: Re: Use "ropes" for string concatenation
From: laruence@gmail.com (Xinchen Hui)

--Apple-Mail-CC8DB9E7-BA19-4E33-BAFB-705AB1FCB7FF
Content-Type: text/plain;
	charset=us-ascii
Content-Transfer-Encoding: quoted-printable

Hey

Sent from my iPhone

> On Mar 24, 2015, at 5:11 PM, Nikita Popov <nikita.ppv@gmail.com> wrote:
>=20
>> On Mon, Mar 23, 2015 at 11:41 PM, Dmitry Stogov <dmitry@zend.com> wrote:
>> Hi,
>>=20
>> Recently, Xinchen and me worked on optimization that eliminates useless r=
eallocations and copying during string concatenation (ZEND_ADD_STRING and fa=
mily + ZEND_CONCAT).
>>=20
>> The idea comes from ropes, but adopted especially for our needs.
>> Rope is popular data structure in languages with immutable strings.
>> http://en.wikipedia.org/wiki/Rope_%28data_structure%29
>>=20
>> We don't try to use ropes everywhere in the engine (at least it's too lat=
er for 7.0), only for concatenation.
>>=20
>> The first branch uses ropes only instead of ZEND_ADD_STRING and family. T=
his must be safe. The only problem is possible memory leaks on exception (bu=
t we already have this problem anyway). The simplest way to understand the p=
atch - read code for new opcodes in zend_vm_def.h.
>>=20
>> https://github.com/php/php-src/pull/1194/files
>>=20
>> The second branch in addition uses ropes for series of ZEND_CONCAT. It di=
sables calls to do_operation(ZEND_CONCAT) handler of custom internal classes=
.
>>=20
>> https://github.com/php/php-src/pull/1195/files
>>=20
>> Both make slight speed improvement (first +0.3%, second +0.6% on wordpres=
s home page).
>>=20
>> We don't currently use ability to override CONCAT behavior in internal cl=
asses, and I'm not sure if it may be useful at all. (For example Lua doesn't=
 provide concat meta-method). May be remove it?
>=20
> Overloading concat operator will be useful in the future for the implement=
ation of the UString class, for example.
>=20
then we can implement is in rope_end handler, and don't convert op to string=
 in rope add var handler.

Thanks
> Nikita

--Apple-Mail-CC8DB9E7-BA19-4E33-BAFB-705AB1FCB7FF--