https://blog.youkuaiyun.com/saint1126zh-cnhttps://blog.youkuaiyun.com/https://blog.youkuaiyun.com/saint1126/article/details/9465135https://blog.youkuaiyun.com/saint1126/article/details/9465135saint1126Thu, 25 Jul 2013 10:31:57 +0800abc, acb, bac, bca, cab, cba六种。 上面我的写法是将首字符在【a,b,c】中循环使用，固定了首字符，次字符在【a，b，c】 - 【{首字符}】中循环，第三位依次推到。看下erlang的递归实现 module(lib_misc). -export([perms/1]). per]]>https://blog.youkuaiyun.com/saint1126/article/details/9464945https://blog.youkuaiyun.com/saint1126/article/details/9464945saint1126Thu, 25 Jul 2013 10:15:33 +0800 []; qsort([Pivot|T]) -> qsort([X|| X <]]>https://blog.youkuaiyun.com/saint1126/article/details/8257941https://blog.youkuaiyun.com/saint1126/article/details/8257941saint1126Tue, 04 Dec 2012 23:19:11 +0800= GREATER > NO_OP no operation 比较器： BinaryComparator  按字节索引顺序比较指定字节数组，采用Bytes.compareTo(byte[])]]>https://blog.youkuaiyun.com/saint1126/article/details/8244507https://blog.youkuaiyun.com/saint1126/article/details/8244507saint1126Fri, 30 Nov 2012 17:57:27 +0800https://blog.youkuaiyun.com/saint1126/article/details/7353846https://blog.youkuaiyun.com/saint1126/article/details/7353846saint1126Wed, 14 Mar 2012 17:17:19 +0800https://blog.youkuaiyun.com/saint1126/article/details/7277573https://blog.youkuaiyun.com/saint1126/article/details/7277573saint1126Mon, 20 Feb 2012 22:25:44 +0800https://blog.youkuaiyun.com/saint1126/article/details/6906670https://blog.youkuaiyun.com/saint1126/article/details/6906670saint1126Wed, 26 Oct 2011 12:31:36 +0800https://blog.youkuaiyun.com/saint1126/article/details/6888719https://blog.youkuaiyun.com/saint1126/article/details/6888719saint1126Wed, 19 Oct 2011 19:06:58 +0800https://blog.youkuaiyun.com/saint1126/article/details/6317029https://blog.youkuaiyun.com/saint1126/article/details/6317029saint1126Tue, 12 Apr 2011 01:24:00 +0800       最近的项目进度比较紧，同步了几个项目，很少做总结，特别喜欢在github上review代码，字体很好，也很清晰。这个项目是实现redis的分布式设计的，之前的一个版本是按照http://www.ooso.net/archives/549 的light cloud的原理实现的，并对get，set做了hset，hget的改进，使得占内存2.4g的key降低到了700M。不过经过每次查询都要经过两次IO，实在是浪费，实测简单的set，get只能达到4000k/s左右，实在是低下。https://blog.youkuaiyun.com/saint1126/article/details/6317012https://blog.youkuaiyun.com/saint1126/article/details/6317012saint1126Tue, 12 Apr 2011 01:01:00 +0800https://blog.youkuaiyun.com/saint1126/article/details/6272169https://blog.youkuaiyun.com/saint1126/article/details/6272169saint1126Wed, 23 Mar 2011 19:57:00 +0800https://blog.youkuaiyun.com/saint1126/article/details/6218946https://blog.youkuaiyun.com/saint1126/article/details/6218946saint1126Wed, 02 Mar 2011 19:07:00 +0800https://blog.youkuaiyun.com/saint1126/article/details/6218349https://blog.youkuaiyun.com/saint1126/article/details/6218349saint1126Wed, 02 Mar 2011 15:41:00 +0800http://www.mjmwired.net/resources/mjm-fedora-f13.html
 ]]>https://blog.youkuaiyun.com/saint1126/article/details/6212305https://blog.youkuaiyun.com/saint1126/article/details/6212305saint1126Mon, 28 Feb 2011 02:25:00 +0800    前几天介绍的mqagent是为memcaheq服务的，写完第二版的蜡笔连接业务后，通过检测memcacheq的队列，发现后端的两台memcacheq的比例严重失调，其中一个都达到了20166355/20166355，而另一个却只有40697/40697。比例500：1.
    虽然memagent中使用了ketama的一致性hash算法，比例却如此失调，看来ketama算法的实现上出现了问题。通过研究一致性hash算法，分别针对libketama库，和memagent自身实现]]>https://blog.youkuaiyun.com/saint1126/article/details/6203834https://blog.youkuaiyun.com/saint1126/article/details/6203834saint1126Wed, 23 Feb 2011 23:51:00 +0800http://labs.alcacoop.it/doku.php?id=articles:redis_land]]>https://blog.youkuaiyun.com/saint1126/article/details/6175601https://blog.youkuaiyun.com/saint1126/article/details/6175601saint1126Wed, 09 Feb 2011 15:32:00 +0800https://blog.youkuaiyun.com/saint1126/article/details/6147169https://blog.youkuaiyun.com/saint1126/article/details/6147169saint1126Mon, 17 Jan 2011 16:06:00 +0800在写蜡笔链接业务的时候和mecacheq的分布式的时候都用到了glib库的hashtable数据结构, 使用了遍历其中所有的键值g_hash_table_get_keys()函数,该函数是glib2.12以后才支持, 服务器上的是glib2.6,版本比较低了,所以打算升级下, wget http://ftp.gnome.org/pub/GNOME/sources/glib/2.26/glib-2.26.1.tar.gz tar xvzf glib-2.26.1.tar.gz cd glib]]>https://blog.youkuaiyun.com/saint1126/article/details/6128824https://blog.youkuaiyun.com/saint1126/article/details/6128824saint1126Tue, 11 Jan 2011 13:02:00 +0800https://blog.youkuaiyun.com/saint1126/article/details/6127779https://blog.youkuaiyun.com/saint1126/article/details/6127779saint1126Mon, 10 Jan 2011 20:33:00 +0800http://www.ibm.com/developerworks/cn/linux/l-cn-signalsec/]]>https://blog.youkuaiyun.com/saint1126/article/details/6119785https://blog.youkuaiyun.com/saint1126/article/details/6119785saint1126Thu, 06 Jan 2011 11:25:00 +0800
看到微博上《程序员杂志》在征集"一分钟先生"的话题：如何做好公司/团队的技术选型？其实大公司或者大一点的团队选型几乎不需要太多讨论的 -- 最后会不可避免的绕到技术官僚的话题上去。这里我想简单说说技术型创业团队技术上的选型问题。
拥抱开源技术
如果只能选择微软的技术路线，比如团队几个人只会用微软的技术做开发，甚至也不想学别的，那么似乎没有别的办法，将就一下吧。如果还有的选择，尽量选择使用开源技术。这样的好处是你不但可以有效的降低软硬件成本，还有更多的部署方案供选择]]>