ViTarn

迅雷 for mac 夹带狗狗搜索

2011-09-09T00:00:00-07:00

迅雷, 你真是不禁夸.

Ruby Koans

2011-07-18T00:00:00-07:00

学习 Ruby 语言的途径有太多太多, 你可以选择一本经典书籍慢慢看, 也可以找在线的资源(类似那种十五分钟入门的), 甚至有很多 Ruby 语言使用者是从 Ruby On Rails 学起的(例如我).

不知道从什么时候开始, Ruby 官网文档里出现一篇Ruby Koans, 采用单元测试填空题的方式强化基础知识. 网站访问不了可以直接访问源代码库.

说真的, 做完这套题目以后, 我对 Ruby 有了重新的认识, 这样说有点太官方了, 其实是我的自信心受到了粉碎性打击然后重组.

将代码下载回来, 运行 ruby path_to_enlightenment.rb, 就会开始单元测试过程, 因为题目的答案还没有填写, 所以测试肯定是通不过的, 碰到问题就会停下来, 然后提醒你哪个文件的第几行出了问题.

最先出现问题的文件会是 about_asserts.rb, 这里是练习一下断言的最基本用法, 以及几道 1+1=2 之类的问题. __处就是要你填空的地方, 如果不填会提醒你 <"FILL ME IN">.

about_nil.rb

#!ruby highlight:5

def test_nil_has_a_few_methods_defined_on_it
    # nil对应空字符串, 所以 "#{nil}" 才不会影响输出.
    assert_equal '', nil.to_s
    # 这个就很新鲜, 之前没有注意到啊.
    assert_equal 'nil', nil.inspect
end

about_objects.rb

#!ruby

def test_some_system_objects_always_have_the_same_id
    # false, true, nil 三者都有固定的 object_id, 为什么会取 0 2 4 这三个值呢? 请看下文.
    assert_equal 0, false.object_id
    assert_equal 2, true.object_id
    assert_equal 4, nil.object_id
end

def test_small_integers_have_fixed_ids
    # 整型的 object_id 等于 2n+1, 所以负数的 object_id 也是负的.
    # 从 0 开始闲置的 object_id 就只有 0 2 4 等等. 所以之前特殊的 object_id 由此得来.
    assert_equal 1, 0.object_id
    assert_equal 3, 1.object_id
    assert_equal 5, 2.object_id
    assert_equal 201, 100.object_id
end

# 其它对象的 object_id 是随机产生并且绝不重复的(包括 Float), object_id 也是判断对象是否为同一个对象的依据.

about_arrays.rb

#!ruby highlight:[8,11]

def test_slicing_arrays
    array = [:peanut, :butter, :and, :jelly]

    # 这种用法对应于 Array#slice(start, length)
    # 当 length 超过数组长度时忽略.
    assert_equal [:and, :jelly], array[2,20]
    # 当 length 为 0 时返回空数组
    assert_equal [], array[4,0]
    assert_equal [], array[4,100]
    # 当 start 超过数组长度时返回 nil.
    assert_equal nil, array[5,0]
end

array[4,0] 返回 [], 而 array[5,0] 却返回 nil. 为什么? Array 文档把这个称为 special cases. 从源代码里顺着 rb_ary_aref 方法继续下去就是 rb_ary_subseq

#!c highlight:[6,14]
rb_ary_subseq(VALUE ary, long beg, long len)
{
    VALUE klass;
    
    // 如果长度大于数组长度, 返回 nil.
    if (beg > RARRAY_LEN(ary)) return Qnil;
    if (beg < 0 || len < 0) return Qnil;

    if (RARRAY_LEN(ary) < len || RARRAY_LEN(ary) < beg + len) {
	len = RARRAY_LEN(ary) - beg;
    }
    klass = rb_obj_class(ary);
    // 如果长度等于 0, 返回空数组.
    if (len == 0) return ary_new(klass, 0);

    return ary_make_partial(ary, klass, beg, len);
}

不知道 Matz 当年是不是晃神了, 还是有意为之, 进而造成了下标 4 和 5 的不一致性, 反正这已经是即定事实了.

about_strings.rb

#!ruby

def test_flexible_quotes_can_handle_multiple_lines
    long_string = %{# \n
It was the best of times,# \n
It was the worst of times.# \n
}
    assert_equal 54, long_string.size
end

def test_here_documents_can_also_handle_multiple_lines
    long_string = <<EOS
It was the best of times,# \n
It was the worst of times.# \n
EOS
    assert_equal 53, long_string.size
end

# += 操作不会改变原始字符串.
def test_plus_equals_also_will_leave_the_original_string_unmodified
    original_string = "Hello, "
    hi = original_string
    there = "World"
    hi += there
    assert_equal "Hello, ", original_string
end

# << 操作会修改源始字符串的.
def test_the_shovel_operator_modifies_the_original_string
    original_string = "Hello, "
    hi = original_string
    there = "World"
    hi << there
    assert_equal "Hello, World", original_string
end

# 单引号字符串里也会发生转义.
def test_single_quotes_sometimes_interpret_escape_characters
    string = '\\\''
    assert_equal 2, string.size
    assert_equal %{\\'}, string
end

def test_you_can_get_a_substring_from_a_string
    string = "Bacon, lettuce and tomato"
    # 可以像数组那样截取字符串分片
    assert_equal 'let', string[7,3]
    assert_equal 'let', string[7..9]
    # 但是如果只取一个字符的话, 返回的是字符的 ASCII 码.
    assert_equal 97, string[1]
end

def test_strings_can_be_split
    string = "Sausage Egg Cheese"
    words = string.split
    # 字符串分割的默认参数是 $;
    assert_equal ["Sausage", "Egg", "Cheese"], words
end

def test_strings_are_not_unique_objects
    a = "a string"
    b = "a string"
    # 值相等.
    assert_equal true, a           == b
    # 但却不是同一个对象.
    assert_equal false, a.object_id == b.object_id
end

这里比较有意思的是 String#split 的默认值 $; . 它是一个全局变量, 它的唯一用途就是作为 split 方法的默认值出现, 它的初始值是 nil. 你可以通过修改这个全局变量来改变 split 方法的行为.

#!bash
>> $; = 'b'
>> "abaabbbccdcbba".split
=> ["a", "aa", "", "", "ccdc", "", "a"]

about_symbols.rb

#!ruby

def test_identical_symbols_are_a_single_internal_object
    symbol1 = :a_symbol
    symbol2 = :a_symbol
    # 符号是独一无二的.
    assert_equal true, symbol1           == symbol2
    assert_equal true, symbol1.object_id == symbol2.object_id
end

# all_symbols 方法可以返回所有符号, 这基中就包括方法名.
def test_method_names_become_symbols
    symbols_as_strings = Symbol.all_symbols.map { |x| x.to_s }
    assert_equal true, symbols_as_strings.include?("test_method_names_become_symbols")
end

# 在 MRI 版本里, 常量也是符号.
in_ruby_version("mri") do
    RubyConstant = "What is the sound of one hand clapping?"
    def test_constants_become_symbols
        all_symbols = Symbol.all_symbols
        
        assert_equal true, all_symbols.include?(:RubyConstant)
    end
end

# 符号可以长的不太一样.
def test_symbols_with_spaces_can_be_built
    symbol = :"cats and dogs"

    assert_equal symbol, symbol.to_sym
end

# 符号也可以在内部插入值.
def test_symbols_with_interpolation_can_be_built
    value = "and"
    symbol = :"cats #{value} dogs"

    assert_equal symbol, "cats and dogs".to_sym
end

# 符号不是"不可变的"字符串, 它没有字符串的方法.
def test_symbols_do_not_have_string_methods
    symbol = :not_a_string
    assert_equal false, symbol.respond_to?(:each_char)
    assert_equal false, symbol.respond_to?(:reverse)
end

# 连最基本的连接操作都不允许.
def test_symbols_cannot_be_concatenated
    assert_raise(NoMethodError) { :cats + :dogs }
end

# 符号可以动态创建, 但是不推荐这样做.
def test_symbols_can_be_dynamically_created
    assert_equal :catsdogs, ("cats" + "dogs").to_sym
end

about_regular_expressions.rb

#!ruby highlight:13

# 字符串可以像这样查找内容.
def test_a_regexp_can_search_a_string_for_matching_content
    assert_equal "match", "some matching content"[/match/]
end

# 找不到会返回 nil, 这种方式更像是条件判断.
def test_a_failed_match_returns_nil
    assert_equal nil, "some matching content"[/missing/]
end

# 带星号的正则表达的十分贪婪.
def test_asterisk_means_zero_or_more
    assert_equal "", "abbcccddddeeeee"[/z*/]
end

为什么会发生的情形? 看一下 Ruby 源代码里都做了什么.

#!c
# 当 String#slice 的第一个参数类型为 Regexp 时, 调用下面这个函数.
rb_str_subpat(VALUE str, VALUE re, VALUE backref)
{
    if (rb_reg_search(re, str, 0, 0) >= 0) {
        VALUE match = rb_backref_get();
        int nth = rb_reg_backref_number(match, backref);
	    return rb_reg_nth_match(nth, match);
    }
    return Qnil;
}

事实上, 是 "abbcccddddeeeee".match(/z*/)[0] 返回了空字符串. 如果换成 "abbcccddddeeeee".match(/missing/) 就返回 nil 了.

#!ruby

# 这会儿又表现的十分容易满足, 这简直就是陷阱.
def test_the_left_most_match_wins
    assert_equal "a", "abbccc az"[/az*/]
end

此处的表现十分怪异, 事实上 Ruby 正则也是遵从正常的贪婪原则的.

#!bash

>> "abb".match /ab*/
=> #<MatchData "abb">
>> "abb".match /ab*?/
=> #<MatchData "a">
>> "abb".match /ab+?/
=> #<MatchData "ab">
>> "abb".match /ab{0,9}/
=> #<MatchData "abb">

# 只是当出现在字符串末尾时就犯糊涂
>> "aabb".match /ab*/
=> #<MatchData "a">
>> "aabb".match /ab{0,9}/
=> #<MatchData "a">

看来以后星号正则要慎重使用了.

#!ruby

# \w 里的连字符是下划线, 不包括减号.
def test_slash_w_is_a_shortcut_for_a_word_character_class
    assert_equal "variable_1", "variable_1 = 42"[/[a-zA-Z0-9_]+/]
    assert_equal "variable_1", "variable_1 = 42"[/\w+/]
end

# 不但能匹配, 还能选择返回第几个匹配结果.
def test_parentheses_also_capture_matched_content_by_number
    assert_equal "Gray", "Gray, James"[/(\w+), (\w+)/, 1]
    assert_equal "James", "Gray, James"[/(\w+), (\w+)/, 2]
end

# 每次匹配, 结果都会被记入全局变量中.
def test_variables_can_also_be_used_to_access_captures
    assert_equal "Gray, James", "Name:  Gray, James"[/(\w+), (\w+)/]
    assert_equal "Gray", $1
    assert_equal "James", $2
end

about_constants.rb

#!ruby

class Animal
    LEGS = 4
end

class MyAnimals
    LEGS = 2

    class Bird < Animal
        def legs_in_bird
            LEGS
        end
    end
end

# 嵌套常量赢过继承常量.
def test_who_wins_with_both_nested_and_inherited_constants
    assert_equal 2, MyAnimals::Bird.new.legs_in_bird
end

class MyAnimals::Oyster < Animal
    def legs_in_oyster
      LEGS
    end
end

# 如果定义成这样, 继承变量就赢了.
def test_who_wins_with_explicit_scoping_on_class_definition
    assert_equal 4, MyAnimals::Oyster.new.legs_in_oyster
end

about_exceptions.rb

#!ruby

class MySpecialError < RuntimeError
end

# 自定义异常完整的继承树.
def test_exceptions_inherit_from_Exception
    assert_equal RuntimeError, MySpecialError.ancestors[1]
    assert_equal StandardError, MySpecialError.ancestors[2]
    assert_equal Exception, MySpecialError.ancestors[3]
    assert_equal Object, MySpecialError.ancestors[4]
end

about_blocks.rb

#!ruby

# 块中的代码甚至会给局部变量带来副作用.
def test_block_can_affect_variables_in_the_code_where_they_are_created
    value = :initial_value
    method_with_block { value = :modified_in_a_block }
    assert_equal :modified_in_a_block, value
end

def test_blocks_can_be_assigned_to_variables_and_called_explicitly
    add_one = lambda { |n| n + 1 }
    # 有两种方法调用 lambda 表达式.
    assert_equal 11, add_one.call(10)
    assert_equal 11, add_one[10]
end

about_scope.rb

#!ruby

class AboutScope
    class String
    end
    
    # 当前范围中的类名拥有更高的优先级.
    def test_bare_bones_class_names_assume_the_current_scope
        assert_equal true, AboutScope::String == String
    end
    
    # 当前范围中的类名如果与 Ruby 库中的类名重复, 会造成很多不必要的麻烦.
    def test_nested_string_is_not_the_same_as_the_system_string
        assert_equal false, String == "HI".class
    end
    
    def test_use_the_prefix_scope_operator_to_force_the_global_scope
        assert_equal true, ::String == "HI".class
    end
    
    MyString = ::String
    
    # 类名其实只是常量而已.
    def test_class_names_are_just_constants
        assert_equal true, MyString == ::String
        assert_equal true, MyString == "HI".class
    end
    
    def test_constants_can_be_looked_up_explicitly
        assert_equal true, MyString == AboutScope.const_get("MyString")
    end
end

about_class_methods.rb

#!ruby

class Dog
end

# 实例也可以定义方法
def test_you_can_define_methods_on_individual_objects
    fido = Dog.new
    def fido.wag
        :fidos_wag
    end
    assert_equal :fidos_wag, fido.wag
end

# 这个方法只属于这个实例, 不影响类的其它实例.
def test_other_objects_are_not_affected_by_these_singleton_methods
    fido = Dog.new
    rover = Dog.new
    def fido.wag
        :fidos_wag
    end

    assert_raise(NoMethodError) do
        rover.wag
    end
end

LastExpressionInClassStatement = class Dog; 21; end

# 你总能拿到最后一个表达式的结果.
def test_class_statements_return_the_value_of_their_last_expression
    assert_equal 21, LastExpressionInClassStatement
end

SelfInsideOfClassStatement = class Dog; self; end

# 甚至是 self.
def test_self_while_inside_class_is_class_object_not_instance
    assert_equal true, Dog == SelfInsideOfClassStatement
end

about_to_str.rb

#!ruby

class CanBeTreatedAsString
    def to_s
        "string-like"
    end

    def to_str
        to_s
    end
end

# to_str 有更进一步的作用.
def test_to_str_allows_objects_to_be_treated_as_strings
    assert_equal false, File.exist?(CanBeTreatedAsString.new)
end

抑制 Ruby 警告

2011-07-13T00:00:00-07:00

在 irb 命令行里输入

#!bash
>> X = 1
=> 1
>> X = 2
(irb):28: warning: already initialized constant X
=> 2

你会看到上面那行警告, 已经初始化过常量 X, ruby 是在提醒你常量是不应该被改变的, 不过 ruby 只是在抱怨一句罢了, 实际上 X 的值已经由 1 变成 2 了.

警告是对的, 常量的再次赋值有违原则. 但是人总会碰到情非得已的时候.

ruby 论坛上有篇贴子也提到了这一问题, 沙发搞笑的说忽略事情最简单的方式就是闭上眼睛. 不过五楼是正经人, 给出了解决问题的思路. 顺着思路就想到了这个方法.

ruby 接受 -W 作为参数.

#!bash
ruby -h
-W[level] set warning level; 0=silence, 1=medium, 2=verbose (default)

运行时对应的是全局变量 $-v

nil -W0
false -W1
true -W2

这样事情就变得简单了.

#!ruby
def suppress_warning &block
    old_v = $-v
    $-v = nil
    yield
ensure
    $-v = old_v
end

试试效果

#!bash
>> X = 1
=> 1
>> suppress_warning { X = 2 }
=> 2

嗯, 感觉不错, 你也是不清理不舒服斯基吗?

迅雷苹果内部预览版 1.0.1.42

2011-07-12T00:00:00-07:00

下载地址

支持 Chrome 和 Firefox 扩展, 不支持 Safari.

就算是支持 Chrome 你不能直接去点迅雷专用链接.

右键菜单里有个”使用迅雷下载”.

“新建下载任务”对话框.

主界面. 下载速度有点慢.

一张有趣的组织结构图

2011-06-30T00:00:00-07:00

摘自 36氪

Google+ 一个干净的 Facebook

2011-06-30T00:00:00-07:00

Google+ 就这样蹦出来了, 让我这个对 Facebook 又爱又恨的人不免心动.

Hello Jekyll!

2011-06-21T00:00:00-07:00

原来 Github 还有这功能.

比 toto 还简单的 Jekyll.

由 Github 帮你生成静态化的站点.

目前感觉相当良好, 只是模板和插件还有点玩不转.

Jekyll on Github

Mac OS X Server Samba crash

2010-08-08T00:00:00-07:00

My Mac pro shutdown because power cut last night.

Today. Windows user can ping me by computer name. But they can’t visit my share folder. There are some error in log:

Samba: ERROR: failed to setup guest info.

When I open /etc/smb.conf. I found it’s blank file. OK! So. To fix this:

#!bash
sudo cp -p /etc/smb.conf.template /etc/smb.conf

Then it works.

Objective-C 笔记

2010-07-24T00:00:00-07:00

声明一个类需要两个文件, 接口(.h)和实现(.m).

接口(interface)更像是”声明”, 协议(protocol)才是起到”接口”的作用.

类方法用加号(+), 实例方法用减号(-).

#!objc
+ alloc;
- init;

引用类型带星号(*).

#!objc
(NSString *) name

方法名包括参数名.

#!objc
// 方法名: setWidth:andHeight:   参数: width, height
- (void)setWidth:(float)width andHeight:(float)height;
// 调用写法
[receiver setWidth:1.0 andHeight:2.0];

super是关键字; self是变量名, 可以被覆盖.

“类采纳了某协议”, 翻译过来就是:”类实现了某接口”. “协议采纳另个协议”, 翻译过来就是:”接口继承”.

#!objc
// Formatter采纳了Formatting和Prettifying两个协议 
@interface Formatter : NSObject < Formatting, Prettifying >
@end

// 协议采纳协议
@protocol Paging < Formatting >

// 使用协议类型表示实例变量
id <Formatting> anObject;

属性声明的格式.

#!objc
@property(这里是一些特性, 例如readonly/retain/copy) 类型 名称;

// IBOutlet的意思是把属性暴露给Interface Builder设计器
@property (nonatomic, retain) IBOutlet NSButton *myButton;

如何在XCode中使用Interface Builder设计器里的对象.

在设计器里选中对象, 切换到Connections标签, 从Referencing Outlets里扯一条线连接到Controller上, 然后选择实例变量名.

数据源委托到控制器.

对于需要集合数据类型作为数据源的界面元素, 会把dataSource连接到控制器上, 然后由控制器采纳数据源协议并实现方法. 这与以往所见的大不相同, 如果有两个数据源该怎么办?

关于#pragma. 是用于XCode识别代码的标记, 被编译器忽略.

方法分类:

#!objc
#pragma mark 采纳UIPickerDataSource协议

白空格:

#!objc
#pragma mark -

压制未使用变量警告:

#!objc
#pragma unused(unusedVar)

只有井底之蛙才会觉得 Web inspector 复制了 Firebug

2010-07-23T00:00:00-07:00

一个部落格饶有兴致的介绍Chrome6.0dev中的开发人员工具, 也就是Web Inspector. 博主谦虚的口吻是清晰可见的, 然而急躁的国人仍然在回复中丝毫不客气, 含蓄的骂着”兰州烧饼”, 还说”对Firebug的复制品无爱”.

谁让咱是个爱钻牛角尖的人咧! 我偏要挖一挖历史:

有据可查的Firebug历史: https://addons.mozilla.org/en-US/firefox/addon/1843/versions/?page=2#version-0.2

有据可查的Web Inspector历史: http://webkit.org/blog/41/introducing-the-web-inspector/

看看吧, 相隔4天, 哪里谈得上谁复制谁.

如果偏要说复制品, 倒是可以看看这篇文章: http://antennasoft.net/robcee/2010/05/14/inspector-landing/

Wash cloth or wash brain

2010-07-02T00:00:00-07:00

Look at a good sample:

And another stupids:

A lazy usage for factory girl

2010-07-02T00:00:00-07:00

Factory girl let us product model instance feel luck. But for me(a lazy boy). I wanna write shorter that standard way.

So a… Look code below. I’m lazy to explain.

#!ruby
# spec/factories.rb

class F
	class << self
		{:a => :attributes_for, :b => :build, :c => :create, :d => :define, :s => :stub, :* => :sequence, :+ => :next}.each do |k,v|
			delegate v, :to => Factory
			alias_method k, v
		end
	end
end

F.d :user do |u|
	u.name "foo"
	u.full_name "福娃"
	u.password "0"
end

F.* :foos do |n|
	"foo_#{n}"
end

# spec/models/user_spec.rb

require "spec_helper"

describe User do
	after do
		User.delete_all
	end

	it "should valid" do
		F.b(:user).should be_valid
	end

	it "should save success" do
		F.c(:user).should_not be_new_record
		User.all.should be_one
	end

	it "should product some foos" do
		F.+(:foos).should eql "foo_1"
		(F + :foos).should eql "foo_2"
	end
end

A joke about mate and pirate

2010-06-27T00:00:00-07:00

Textmate has a shell version named “mate”. It’s a simple way to open a project directory.

#!bash
mate ~/Documents/myproj

I use Chinese Wubi input method. And I always forgot switch it. So when I input:

mate

The output word is:

贼船

Translate into English:

Pirate ship

lol

When rails ujs met chrome frame

2010-06-11T00:00:00-07:00

Rails ujs let us use other javascript frameworks (like jQuery) in our project. That’s cool.

Here is an article about top 10 javascript frameworks by google. He use a way (an easy way) to sort these frameworks. Search “javascript frameworks” in Google. You can see that the default rails javascript frameworks prototype is not the most popular one now.

I’m working on a small project for my company by rails and jQuery. The newest rails3 use many html5 features. So I decide let user use a modern browser to use my web page. Otherwise IE6 will kill me. How to switch smoothly? I found chrome frame has beta yesterday!

Someone said that they hate the web page of xxx-browser only. But wait! I use this in my company only! Didn’t you saw those enterprise-level soft let you install their plugin?

Make an IE-only page is easy. Make an IE-except page is easier. lol

Chrome frame show my html5 page (maybe). It’s beautiful than IE. But it cannot works when I send ajax request.

For example I write a link like this:

#!html
<!-- Rails html5 templete -->
<%= link_to "Add user", new_user_path, :remote => true %>

<!-- Will product -->
<a href="/users/new" data-remote="true">Add user</a>

Rails handle these link with data-remote by rails.js, It cancel click event, make a ajax request, then trigger ajax:success event back.

#!javascript
// Live is new method in jQuery 1.4+.
$('a[data-confirm]').live('click', function () { ... });

// When user click the link. jQuery will handle it and send a ajax request.
$.ajax({
	url: "/users/new",
	dataType: "script",
	...
});

But chrome frame look like send a incorrect request to rails. Check the log of rails, I found this line:

Processing by UsersController#new as */*

The right process method is JS because the data type is script. I have test IE FF Chrome and Safari on win and mac. Only chrome frame has this issus.

I think I should fix this at client side. When I try to get navigator.userAgent by javascript. I found it has the same value as chrome stand-alone.

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/533.4 (KHTML, like Gecko) Chrome/5.0.375.62 Safari/533.4

But rails could detect the right user agent:

Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 2.0.50727) chromeframe/5.0.375.62

Oh! Javascript don’t know it’s running in chrome frame, so it can do anything. Rails know it’s processing the request from chrome frame, but it don’t know that’s a ajax request.

After a minute. I decide let rails tell javascript: “u r working for chrome frame.”

application.html.erb

#!html
<!-- Hold it in application.html.erb or ur own default layout. for this example put in navigator. -->
<% if @win_ie_with_chromeframe %>
<script>
navigator.chromeframe = true;
</script>
<% end -%>

application_controller.rb

#!ruby
# Detect chrome frame in application_controller.rb
before_filter :chromeframe_detect

protected
def chromeframe_detect
  ua = request.user_agent
  @win_ie_with_chromeframe = ua =~ /Windows NT/ and ua =~ /MSIE (\d+)/ and $1.to_i < 9 and ua =~ /chromeframe/
end

And then hack $.ajax. In fact I have three things need fix.

Chrome frame cannot send PUT DELETE verb.
Append .js to url tell rails I’m a JS request.
jQuery will try parse the result as JSON when dataType is script. If it got html. You will see an error: “Uncaught SyntaxError: Unexpected token <”.

application.js

#!javascript
//BUGFIX: Chromeframe miss some features.
(function($) {
	$.oldAjax = $.ajax;

	$.ajax = function(o) {
		// I can't detect it by navigator.userAgent. So I set it on server side.
		if (navigator.chromeframe) {
			// Chromeframe can't send DELETE and PUT verb.
			if (o.type == "DELETE") {
				o.type = "POST";
				o.data = "_method=delete";
			} else if (o.type == "PUT") {
				o.type = "POST";
				o.data += "&_method=put";
			}
		}

		// Chromeframe send incorrect ajax type make rails process the request by */*. The right way is JS. I try many way to avoid this. But I can do this with append .js to url only.
		if (o.dataType == "script" && !/\.js$/.test(o.url)) {
			o.url += ".js";
			// Webkit try parse result to json then raise a syntax error when receive html. It appear in inspector only and user can't see that except u r a developer. The url(end of .js) has told rails how to process my request. So I set dataType as text because I need respondText only. I can parse json myself.
			o.dataType = "text";
		}

		$.oldAjax(o);
	};
});

貌似 AutoCAD 又回到 Mac 了

2010-05-23T00:00:00-07:00

Adobe Air2rc1 drag List item to Trash or RecycleBin has different behavior

2010-05-15T00:00:00-07:00

#!xml
<s:List dragEnabled="true" dragMoveEnabled="true">
	<s:dataProvider>
		<s:ArrayCollection>
			<fx:Object label="Item one"/>
			<fx:Object label="Item two"/>
			<fx:Object label="Item three"/>
		</s:ArrayCollection>
	</s:dataProvider>
</s:List>

I think this is a bug. But I have no solution.

Mac 服务器 servermgrd 崩溃

2010-05-12T00:00:00-07:00

首先说一下背景:

我们平常用Server Admin来管理MacOSX Server时, 是通过图形界面与一个叫servermgrd的后台进程沟通交换数据的, 它由LaunchCtl启动/System/Library/LaunchDaemon/com.apple.servermgrd.plist实现开机自动运行, 侦听端口311.

事实上servermgrd是一个Web服务器, 我们的Server Admin是通过HTTP的方式实现远程管理的(即便你在本机使用). 你可以在本机上打开https://localhost:311印证这一点.

不得不说servermgrd的启动依赖太多东西, 修复的过程让我意识到许多服务的配置都会间接的导致servermgrd崩溃. 我认为这会导致许多不便, 问题的根源是servermgrd不能尝试去修复你的配置, 但至少应该兼容这些未包括在图形界面上或不正确的配置, 例如当配置Apache的时候用户自己打开配置文件加入的参数Server Admin是不会碰的. 总之, 不应该崩溃, 因为它一旦崩溃, Server Admin就无法修改, 这真的很尴尬.

我的Mac Pro运行着Mac OS X Server 10.6.3. 在一次运行Gateway Setup Assistant失败后, 遭遇了servermgrd不能启动的尴尬局面. Console里显示servermgrd不能启动的原因是”Segmentation fault”.

#!plain
May 12 14:20:39 mes servermgrd[7658]: servermgr_ipfilter:ipfw config:Notice:Flushed IPv4 rules
May 12 14:20:39 mes servermgrd[7658]: servermgr_ipfilter:ipfw config:Notice:Flushed IPv6 rules
May 12 14:20:39 mes com.apple.launchd[1] (com.apple.servermgrd[7658]): Job appears to have crashed: Segmentation fault
May 12 14:20:39 mes com.apple.launchd[1] (com.apple.servermgrd): Throttling respawn: Will start in 9 seconds
May 12 14:20:39 mes com.apple.ReportCrash.Root[7607]: 2010-05-12 14:20:39.824 ReportCrash[7607:440b] Saved crash report for servermgrd[7658] version ??? (???) to /Library/Logs/DiagnosticReports/servermgrd_2010-05-12-142039_localhost.crash

我尝试着google各种可能的关键词, 发现servermgrd不能启动的原因真的太多了. 许多问题都没有答案, 解决办法往往是重装系统.

没有了Server Admin这个图形化的工具, 其实我们还有serveradmin这个shell工具, 它不需要与后端通讯, 只是用起来有点不方便.

我是在运行向导时出现的问题, 因此我将目标锁定在NAT服务. (说起来容易, 其实到此为止我已经浪费了大半天的时间了.)

#!bash
# 查看NAT服务配置
sudo serveradmin settings nat
nat:reverse = no
nat:unregistered_only = yes
nat:same_ports = yes
nat:natportmap_interface = "en0"
nat:log = yes
nat:clamp_mss = yes
nat:dynamic = yes
nat:log_denied = no
nat:use_sockets = yes
nat:proxy_only = no
nat:enable_natportmap = yes
nat:interface = ""
nat:deny_incoming = no

看到了没? NAT的接口竟然指向了空字符串.

nat:interface = ""

修复方法:

#!bash
# 试着改回正常值
sudo serveradmin settings nat:natportmap_interface = "en1"
nat:natportmap_interface = "en1"
sudo serveradmin settings nat:interface = "en0"
nat:interface = "en0"

# 等10秒看看servermgrd起来没
ps aux | grep servermgrd | grep -v grep
root      7687   0.0  0.2  2577788  38484   ??  Ss    2:20PM   0:03.57 servermgrd -x

顺便跟着乔老爷子起哄, BS一下Adobe.

起死回生的Server Admin再一次尝试连接时, 会出奇的慢, 似乎在重置着自己的配置. 等了N久后显示 Server Admin wants to use the PrivateEncryptedDatak keychain, 并且提示我输入密码, 可是怎么输也不对.

你有没有觉得 PrivateEncryptedDatak 很眼生, 它是谁呀? Google后发现它是Adobe Air的一个示例程序生成的Keychain. Adobe suck!

删了它, 没事的, 反正就一Demo.

位置: ~/Library/Application Support/Adobe/AIR/ELS/com.adobe.demo.RememberMe/

如果你有洁癖, 就继续打开Keychain - Edit - Keychain List. 把那里面的也删除了.

Google 真是越来越可爱了

2010-04-26T00:00:00-07:00

Google 真是越来越可爱了, 看图.

循环输入用于 Google 日历的 .ics 格式文件

2010-04-16T00:00:00-07:00

#!ruby
require 'date';

start, days, summary = DateTime.new(2010,1,1), 76, 'åˆ°';

puts %{
    BEGIN:VCALENDAR
    PRODID:-//Google Inc//Google Calendar 70.9054//EN
    VERSION:2.0
    CALSCALE:GREGORIAN
    METHOD:PUBLISH
    X-WR-TIMEZONE:UTC
    X-WR-CALDESC:
};

days.times {
    puts %{
        BEGIN:VEVENT
        DTSTART;VALUE=DATE:#{start.strftime('%Y%m%d')}
        DTEND;VALUE=DATE:#{start = start.next; start.strftime('%Y%m%d')}
        SUMMARY:#{summary}
        END:VEVENT
    }
};

puts %{END:VCALENDAR}

Hello Toto!

2010-04-08T00:00:00-07:00

About:

Write artile by markdown and publish by git.

Why?

Why not?

Do u like do a simple thing (write) with a complex way?

I like toto. Because it so simple and cool. What’s blog title? What’s category or tags? What’s pingback and comment? What’s admin control panel? Whatever???

Support

中文

1号字

2号字

3号字

4号字

5号字

6号字

列表
- 子项
  - 子子项

粗体

斜体

链接