你不需要知道的Ruby实现

你不需要知道的

各種Ruby實現

關於我

莫樞 Kris Mok

• 目前在淘寶就職，參與後台系統的開發。
• 對編程語言的設計與實現非常感興趣，
• 主要在編譯器與虚擬機方面
• 接觸Ruby開始就對其愛不釋手

希望能與大家共同學習進 ^_^

• 博客：

http://rednaxelafx.javaeye.com

• 高級語言虚擬機圈子

http://hllvm.group.javaeye.com

@rednaxelafx

毎種實現都有自己的特點

各自都有獨特的擴展

不要擔心“社區分裂”“標準不一”

核心語言都是一樣的

• RubySpec
• Ruby語言規範

但種實現通過獨特的擴展能為各自的場景提供最好的服務

1 + 2 * 3

Ruby 1.8: MRI

• 遍抽象語法樹的解釋器
• 保守式、標記-清掃式GC
• 綠色線程

irb(main):001:0> ISeq = RubyVM::InstructionSequence

=> RubyVM::InstructionSequence

irb(main):002:0> iseq = ISeq.compile '1 + 2 * 3'

=> <RubyVM::InstructionSequence:<compiled>@<compiled>>

irb(main):003:0> puts iseq.disassemble

== disasm: <RubyVM::InstructionSequence:<compiled>@<compiled>>==

0000 trace 1 ( 1)

0002 putobject 1

0004 putobject 2

0006 putobject 3

0008 opt_mult

0009 opt_plus

0010 leave

=> nil

Ruby 1.9: YARV

• “字節碼”解釋器
• 保守式、標記-清掃式GC
• 操作系統的原生線程或綠色線程
• 有GVL
• 改良的正則表達式引擎

主要用C語言實現

Re: Ruby GC question (MRI, JRuby, etc)

Posted by Yukihiro Matsumoto (Guest) on 2010-08-21 02:15

Hi,

In message "Re: Ruby GC question (MRI, JRuby, etc)"

on Sat, 21 Aug 2010 08:15:15 +0900, Jörg W Mittag

<JoergWMittag+Ruby@GoogleMail.Com> writes:

|In particular, the GC algorithms in MRI and YARV are specifically

|designed with the assumption that they will never actually run in

|99.999% of all cases. They are designed for scripting, where a script

|doesn't even allocate enough memory to trigger a collection, runs for

|a couple of seconds and then exits, after which the OS simply reclaims

|the memory: no GC needed.

99.999% is a bit over-exaggerated, but it is true that garbage

collection algorithm of YARV and MRI focus for throughput on

non-memory extensive short-running programs, and GC of REE is not

suitable for those programs.

matz.

IronRuby

• 建立于DLR与CLI平台上
• 充分利用底層平台已有的功能，如解釋器/JIT編譯器/GC/線程
• 与.NET/Silverlight

/Mono/Moonlight無縫整合

• 兼容1.8.6（1.0/1.1）与1.9.2（1.1.1）

主要用C#語言實現

Where is my Ruby?

Rubinius

• 全新的Ruby實現
• 兼容1.8.7
• 先進的虚擬機
• 解釋器與JIT編譯器組成的混合執行模式
• JIT編譯器基於LLVM
• 先進的GC
• 本地線程
• 即將削除GIL
• 強大的調試器

字節碼編譯器與Ruby核心庫

與標準庫都主要用Ruby實現

可用于帮助理解Ruby核心的行為

class Array

include Enumerable

# Passes each index of the Array to the given block

# and returns self. We re-evaluate @total each time

# through the loop in case the array has changed.

def each_index

return to_enum(:each_index) unless block_given?

i = @start

total = i + @total

while i < total

yield i

i += 1

end

self

end

end

static VALUE

rb_ary_each_index(VALUE ary)

{

long i;

RETURN_ENUMERATOR(ary, 0, 0);

for (i=0; i<RARRAY_LEN(ary); i++) {

rb_yield(LONG2NUM(i));

}

return ary;

}

void

Init_Array(void)

{

rb_cArray = rb_define_class("Array", rb_cObject);

rb_include_module(rb_cArray, rb_mEnumerable);

rb_define_method(rb_cArray, "each_index", rb_ary_each_index, 0);

}

class Fixnum < Integer

def +(o)

Ruby.primitive :fixnum_add

redo_coerced :+, o

end

end

class Fixnum : public Integer {

public:

// Ruby.primitive! :fixnum_add

Integer* add(STATE, Fixnum* other) {

native_int r = to_native() + other->to_native();

if(r > FIXNUM_MAX || r < FIXNUM_MIN) {

return Bignum::from(state, r);

} else {

return Fixnum::from(r);

}

}

}

有多先進？

114.8!

Rubinius 1.0RC1的速度是

Ruby 1.8的114.8倍！

• 在什麼測試上？

為什麼能快那麼多？

JIT編譯器的優化

方法内聯

method inlining

無用代碼削除

dead code elimination

先進的JIT編譯器

提升Ruby代碼性能

Rubinius内核有大量Ruby

因此JIT編譯器帶來

整體平台的性能提升

還有先進的GC

分代式GC

降低GC造成的應用暫停時間

年輕代：成熟的Baker GC

年老代：新鋭的Immix GC

2007年論文：

Immix Garbage Collection:

Fast Collection, Space Efficiency,

and Mutator Locality

module GC

# The Maglev in-memory garbage collector runs

# automatically whenever GC is required.

# It cannot be disabled.

def self.start

# has no effect

end

def self.enable

false # gc is never disabled

end

def self.disable

# has no effect

false

end

def garbage_collect

# has no effect

end

def self.stress# added for 1.8.7

false

end

def self.stress=(bool)# added for 1.8.7

# has no effect

false

end

end

大家熟悉和喜愛的

果咩那塞 >_<

Ruby 1.8: MRI

• 遍抽象語法樹的解釋器
• 保守式、標記-清掃式GC
• 綠色線程

RubyConf 2009上的Rubinius session

(Evan Phoenix)

Ruby程序員的Ruby！

module GC

def self.start

run(false)

end

def self.run(force)

Ruby.primitive :vm_gc_start

raise PrimitiveFailure, "GC.run primitive failed"

end

# Totally fake.

def self.stress

@stress_level ||= false

end

# Totally fake.

def self.stress=(flag)

@stress_level = !!flag

end

# Totally fake.

@enabled = true

def self.enable

# We don't support disable, so sure! enabled!

ret = !@enabled

@enabled = true

return ret

end

# Totally fake.

def self.disable

# Treat this like a request that we don't honor.

ret = !@enabled

@enabled = false

return ret

end

def garbage_collect

GC.start

end

end

module ObjectSpace

def self.find_object(query, callable)

Ruby.primitive :vm_find_object

raise PrimitiveFailure, "vm_each_object failed"

end

end

static VALUE

fix_plus(VALUE x, VALUE y)

{

if (FIXNUM_P(y)) {

long a, b, c;

VALUE r;

a = FIX2LONG(x);

b = FIX2LONG(y);

c = a + b;

r = LONG2NUM(c);

return r;

}

switch (TYPE(y)) {

case T_BIGNUM:

return rb_big_plus(y, x);

case T_FLOAT:

return DBL2NUM((double)FIX2LONG(x) + RFLOAT_VALUE(y));

default:

return rb_num_coerce_bin(x, y, '+');

}

}

void

Init_Numeric(void)

{

rb_cNumeric = rb_define_class("Numeric", rb_cObject);

rb_cInteger = rb_define_class("Integer", rb_cNumeric);

rb_cFixnum = rb_define_class("Fixnum", rb_cInteger);

rb_define_method(rb_cFixnum, "+", fix_plus, 1);

}

typo: it's :* here

What you don't need to know

about Ruby implementations