[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8},"actionData","errors","root",{"_11":-5},"routes/user.$name.replies",{"_9":10,"_11":12,"_13":14,"_15":16,"_17":18,"_19":16},"replies",[35],"user",{"_20":21,"_13":14,"_22":23,"_24":25,"_26":27,"_28":29,"_30":31,"_32":16,"_33":34},"loginname","fisherman","page",1,"limit",50,"total","id","586","avatar_url","http://www.gravatar.com/avatar/e897dd019213ce32752c90567769ccec?s=48","githubUsername","","create_at","2011-09-06 22:22:09","score",5,"topic_count",0,"reply_count","is_star",false,{"_20":36,"_37":38,"_39":-5,"_40":41,"_42":43,"_44":45,"_46":34,"_47":34,"_32":48,"_49":50,"_26":51,"_52":53},"43","author_id","277","tab","content","由Ryan的邮件可知，node v0.8分支已经开始开发，预计将于一月初发布。v0.8分支将带来如下的新特性：\r\n
\r\n
\r\n
\r\n
1. 将src/platform-*.cc中所有平台相关的函数迁移至libuv中。\r\n
\r\n
2. 移除WAF（目前node在*nix下使用的编译工具），以后所有平台下都将使用GYP来编译。GYP是由Chromium项目引入的编译工具。\r\n
\r\n
3. 目前V8通过Isolate，可以实现在同一进程的多个线程中执行独立的Javascript代码。node将对这一特性提供直接支持，即允许用户在新线程中执行Javascript代码。详情将在后文叙述。\r\n
\r\n
4. 提供一种叫做Domains的机制，将与不同网络连接相关的I/O操作分离开来。当一个domain出现了异常时，domain内所有关联的I/O handle都会被清理掉。比如来了一个HTTP请求，然后这个请求发生了异常，所有和这个请求相关的I/O都会被清除。\r\n
\r\n
5. node目前使用了一个内存池来分配数据流使用的buffer内存，但是这个内存分配器在很多时候效率并不高，对此将进行改进。\r\n
\r\n
6. 将提供更好用的编译C++ extension的工具，以便于为所有平台开发extension。\r\n

Isolate

\r\n
这几条中，最令人关注的是第3点。node在未来将提供多线程支持吗？某种程度上可以说是。\r\n
\r\n
node将要提供的Isolate接口，和child_process的spawn非常相似，即在完全分离的环境里执行代码，但在Javascript层面上并不共享内存和代码。\r\n
\r\n
但不同的是，每个Isolate都处在独立的线程中，而child_process则是处在独立的进程中。前者在C层面上是可以共享内存的，通过编写C++模块来提供接口，我们就可以实现在不同Isolate间共享同样的资源。用node进行方便的并行操作也将成为可能。\r\n
\r\n
为了实现这个特性，node首先需要让自己更适合多线程环境，所有的全局变量都将被移除，大量的debugging也避免不了。对于编译的C++模块，它们也需要判断自已要为哪个Isolate服务，而且，一个模块到底需要被加载几次，也变得难以判断起来。\r\n
\r\n
同时，当一个Isolate结束时，整个进程并不会结束，node还需要结束掉和这个Isolate相关的I/O，这一点主要通过domains来实现（见上面第四条）。\r\n
\r\n
所以，期待2012吧。","title","node v0.8 Roadmap详解","last_reply_at","2012-05-15 04:56:15.759","good","top",12,"visit_count",10672,"2011-11-19 13:38:46","author",{"_13":54,"_22":55},"fool","http://www.gravatar.com/avatar/dc13d329db98a4ff44d52013b80e3bf4?s=48"]