1. Apache优化

ubuntu下有个apache2-mpm-perfork包，能很显著地降低apache内存占用。装完之后apache的从200M掉到100M！很好很强大！

2. OpenVPN证书生成

这个要在本机做！生成的CA.crt是public key,ca.key是private key，这两个用于生成server和client的证书。因此ca.key一定不要传到server上去，否则别人就能拿来自己生成证书了。

3. 服务器内核配置

服务器端要做nat的话需要内核的tun和iptables_nat module，我的vps是联系管理员给加的，因为自己改不了内核配置。

4. OpenVPN服务器端配置

做NAT需要添加两条：
这个将使客户端将VPN配置成默认网络连接

push "redirect-gateway def1"

这个将设置客户端的DNS服务器（防止DNS劫持）

push "dhcp-option DNS x.x.x.x"

这里DNS可以用OpenDNS的server

5. 用iptables中转来自client的包

iptables -t nat -A POSTROUTING -s 10.8.0.0/24 -o eth0 -j SNAT --to-source x.x.x.x

–to-source配置成服务器公网IP。OpenVPN的howto里面的这一条目在我的VPS上不适用。

6. Windows客户端权限问题

windows的GUI客户端需要管理员权限运行。在vista/win7等开启了UAC时，要将gui程序兼容性改为winxp,并默认管理员运行。

7. 客户端route/iptables

按上面配置客户端将会使用VPN做默认连接。可以使用chnroutes来配置国内IP不走VPN.

客户端配置仍然不太满意。正在考虑Firefox的autoproxy插件，再利用什么软件中转到VPN去，然后其余连接用普通的。这样可以大大减少VPN流量并提高访问速度。刚在google code中搜了下openvpn，似乎有相关的项目。firefox那边插件选择连接的可能性不大，理论上。
客户端都配好了再更新。

—PS—
今天本来想跟美女炫耀一下，结果说错话了，适得其反。郁闷。

Also this will be fixed in eclipse 3.5.2, the is a work around for those older 3.5 releases. Write a simple shell script to start eclipse:

#!/bin/sh
export GDK_NATIVE_WINDOWS=1
~/bin/eclipse

Replace "~/bin/eclipse" with your eclipse path.

是在大概两周之前提交了备案，昨天刚刚通过了，然后今天就马上整理好东西重新开通了。看看荒芜的blog，倒也有些惭愧。年前在玩python，后来因为unicode的问题放弃了。年后搞到xbox之后就不务正业啦，下班之后的时间几乎全耗在上面了。再之后一些残念，一一作古。现在游戏玩腻了，也许该重新整整网站了。重新开张，重新写东西，算作第一步吧！

放个鞭炮吧，噼啪噼啪，噼啪噼啪…

My website gets a high pagerank. It was 4 about 2 months ago. But now drop back to 3. I don’t know but maybe they choose to spam on sites with high pagerank. Thanks for Akismet, there were only about 2 or 3 spam comments wasn’t blocked. So the spam links never get a link to promot there page rank.

I want to put it in sidebar. So I added a widget function for this plugin. Also the content filters were commented out so that it won’t borther to filter my post contents everytime since I don’t want to display them in post or feeds.

Although it works fine for me now, there are still some things to be improved:

1. If the number of related posts does not reach to display limit, it should have option to fill with random or most commented posts.
2. The title section really need to be moved out from the main code generate function.
3. Take count of category. The posts in the same category are related.
4. Denis submitted v0.6 to SVN trunk! I wander how he released v0.7 via SVN…

I will send Denis my patch tomorrow. Hopefully he will merge it into next release, and, fix the functions I wanted here.

基本功能是这样的: 在地址栏输入sandbox能访问我建在原来localhost下的document root/sandbox文件夹.

分开来,包括了两大部分: 将sandbox解析到本机(http request给Apache); 然后Apache根据server name返回内容.

解析sandbox

当你往浏览器地址栏里面输入http://sandbox/的时候, 浏览器会把sandbox作为一个域名, 查询与之相关联的IP. 浏览器首先去查看系统的host文件有没有列出这个域名. 如果有, 将直接返回host文件里的IP. 没有, 则要找网络连接里面配置的DNS服务器查询IP.

我要达到的目的让浏览器查询sandbox时返回本机IP 127.0.0.1, 最简单的办法就是修改host文件啦.

host文件在我的系统(Vista)的C:\Windows\System32\drivers\etc内. 用记事本打开, 在文件末端添加一行:

127.0.0.1    sandbox

这样浏览器就会将sandbox指向127.0.0.1.

配置Apache

浏览器把http请求顺利的转到了本机, 接下来应该是Apache忙活了. 为了让Apache找到sandbox对应的本地目录, 就需要配置一个VirtualHost.  我的配置是这样的:

NameVirtualHost 127.0.0.1:80
<VirtualHost 127.0.0.1:80>
DocumentRoot "F:/web"
ServerName localhost
</VirtualHost>

<VirtualHost 127.0.0.1:80>
DocumentRoot "F:/web/sandbox"
ServerName sandbox
</VirtualHost>

具体配置说明去看文档, 这里就不讲了. 上面配置的意思就是127.0.0.1下域名为localhost的内容在F:/web, 而sandbox内容在F:/web/sandbox.

到这里, 浏览器就能收到Apache的输出, 显示存在F:/web/sandbox下的内容了.

未解决的问题

host文件似乎不能用将域名绑定到一个端口上去. 比如我尝试下面的配置就失败了:

127.0.0.1:8000    jira

http://jira/不会连接到我的Tomcat上, 而是很”智能”地跑到jira.com去了-_-!!!也许这种问题需要其它的解决方案咯.

1. It’s pure XML, which is similar to XHTML2.
2. It is supported by current common browsers.

Using XHTML1.1 will let you be familiar with pure XML, and makes it much easier to change to XHTML2 in the future.

I decided to switch to XHTML1.1 (from XHTML1.0 traditional) today. While cleaning up my web pages, I found some tips which might be helpful for web you developers:

Doctype declare:

This is the first thing to do. Add following line in the very beginning of your file:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
"http://www.w3.org/TR/2001/REC-xhtml11-20010531/DTD/xhtml11-flat.dtd">

This will let browsers know it’s in XHTML1.1 format.

Also, because XHTML1.1 should be transmitted with MIME type application/xhtml+xml, a meta tag should be added in <head>:

<meta http-equiv="Content-Type"
    content="application/xhtml+xml; charset=utf-8" />

No lang attribute

“lang” attribute in tags is not supported in XHTML1.1. Instead, you should use “xml:lang” attribute. A example:

<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">

This defines language of the whole document as English(en).

Other changes from XHTML1.0 strict can be found here: http://www.w3.org/TR/xhtml11/changes.html#a_changes

Adding CDATA in scripts

Some java scripts may break validation result of your web page. A very common example is Google Analytics javascript code. Modify your Google analytics code will make it passing W3C validation:

<!--Google Analytics-->
<script type="text/javascript">
//<!--[CDATA[
var gaJsHost = ( ("https:" == document.location.protocol) ?
    "https://ssl." : "http://www.");
document.write("\<script src='" + gaJsHost +
    "google-analytics.com/ga.js' type='text/javascript'>\<\/script>" );
//]]-->
</script>
<script type="text/javascript">
//<!--[CDATA[
var pageTracker = _gat._getTracker("UA-????????");
pageTracker._initData();
pageTracker._trackPageview();
//]]-->
</script>
<!--End of Google Analytics-->

File name matters

Traditionally, it does not matter if you name web pages with .hml/.html. But if you still use them as postfix of XHTML1.1 web pages, your server most probably send it withMIME type text/html while XHTML1.1 should be application/xhtml+xml. Does this matter? W3C validator warning message says:

Using a wrong media type for a certain document type may confuse the validator and other user agents with respect to the nature of the document, and you may get some erroneous validation errors.

I tried with my webserver(FreeBSD+Apache), it will send right mime type if file name ends with .xhtml. Otherwise W3C validator will generate a warning message. But unfortunately my server won’t recognize index.xhtml as default page. So be sure to check this issue on your server. Otherwise your front page might be broken.

Although I would like to suggest everyone to drop his IE6, it’s disappointing for web developers to see that IE6 is still widely used. And IE6 has problem with MIME typeapplication/xhtml+xml as described here. This might cause problem under some conditions – especially when you change file postfix to .xhtml. So it seems to be a hard decision to make on file names – although it’s easy for me: I abandon users under IE6!

Setting MIME type

Setting MIME type in dynamic language is easy. Here is a script in PHP:

header("Vary: Accept");
if (stristr($_SERVER["HTTP_ACCEPT"], "application/xhtml+xml"))
header("Content-Type: application/xhtml+xml; charset=utf-8");
else
header("Content-Type: text/html; charset=utf-8");

This script will check with user agent to see if it accept application/xhtml+xml, thus it’s safe to use.

日程上有一个issue是写一个JIRA的安装教程, 还有一些新鲜的东东, 比如M$和DELL搞的(product)RED, android下的bluetooth, HID, 还有一些新鲜的ideas, 崔健的Rock, 包括刚刚从taobao淘到的黑曜石手链和Diesel jeans… 太多太多

ADC剩下的时间不多了, 能用的idea才刚刚出来, 看来最近也不可能抽时间写blog了. 再过半个月, 又要忙工作的事了… 算来算去, 要熬到哪天才能安心的写篇blog呢? 唉, 50h/d的幻想又一次浮出脑门~~~

Unfortunately because of GFW, I can’t see that live demo video. But from the picture I can see it’s running the newly released m5 build of android OS. I think the release of m5 is based on the test on that mobile phone – the screen size and menu layout in SDK m5 is the same as this phone.

So, it’s coming soon~~~ I gonna hurry up with my ADC project.

不得已,再去翻wp-db.php. 却发现代码改了,添加了对DB_CHARSET的支持,使用了:

$this->charset = DB_CHARSET;
...
$this->query("SET NAMES '$this->charset'");

来定义数据库连接的编码了

于是马上在我的wp-config.php中添加了一行:

define('DB_CHARSET','UTF8');

F5了一下, 果然好用, 哈哈．以后不需要改wp的core, 直接在config文件中定义就可以了. 方便,方便,相当地方便.