HTML中相对URL的解释
1、真正的相对路径举个例子，假如你有一个调试用的服务器地址为"http://localhost"。其下还有几个子功能。包括http://localhost/test1/index.html和http://localhost/test1/hello.html。
这时就可以考虑使用相对路径了。假如在index.html页面中要链接到hello.html页面去。可以在链接标签中使用href="hello.html"就可以正确的链接过去。因为index.html页的当前的目录是/test1，所以相对于index.html的链接就转换为/test1/hello.html。
2、全局相对路径其实这一段的所谓相对路径，仅仅是指不含有主机名的路径而已。
主要特征就是在相对URL的最前面是含有"/"的。这样的路径就是以网站的根路径开始计算的。
还是第一种方法中的例子，换成全局相对路径可以写成 /test/hello.html 的形式；或者index页面想要引用 http://localhost/css/style.css,按照1的写法可以写成 <link href="../css/style.css" />，虽然还是看的懂，但总不是那么简洁。
所以可以考虑使用全局相对路径， 就可以写成这样的形式<link href="/css/style.css" />，这样是以URL的根路径开始寻找的，可以很好的找到需要的路径。
全局相对路径是一种很好的方式，可以实现在一个WEB应用的部署内部很好的链接，也不需要为了".."而晕倒。问题就是，如果一个应用的部署从"/"变成了"/app"之类的，就会出问题了。而这种情况在Tomcat等等不太适合在根路径上配置服务的服务器就比较常见了。
3、框架中的相对URL 框架中的相对URL就比较容易理解了。比如在一个框架中制定另一个框架页面需要打开的相对URL，这时这个相对URL就是基于本框架页面的。
比如左侧框架页面正在显示的页面是/hello/test.html，而右侧框架页面正在显示/hi/test.html。则在左侧框架页面的一个连接<a href="test2.html" target="XXX"></a> 在右侧框架页面打开连接的话，则是打开/hello/test2.html。
4、框架中的JavaScript的相对这个就比较恶心了，在左侧框架页面中的一个连接上使用事件来处理JS代码。JS代码中直接引用右侧框架页面对象，并指定相对URL，则这时的相对URL是基于原来的目的框架页面的。
在这种情况中是很容易与第3条相混淆的，应该注意。
这里是google的工程师在论坛里对相对路径和绝对路径的解答，我觉得不错，引用下来。
There is no "absolute" answer :-) There are pros and cons to both
absolute and relative URLs in links: Absolute URLs: + help keep the
links pointing to your content if someone were to copy it (*) + help
keep the links pointing to your domain name if you cannot select a
canonical (can't do 301 redirects) + help make sure that you're
pointing to the right URL even if you move things around (say for
stylesheets or graphics) - cannot be tested on a staging / testing
server (eg locally) (unless you insert the links dynamically) - makes
it hard to move content (unless the links are inserted dynamically)
Relative URLs: + make it easy to move content around + make it easy to
test locally and on a staging server - are easy to break if linking to
content that isn't moved as well (stylesheet, graphics, etc) - an evil
scraper would have less work (*) There's a middle ground as well,
using absolute links without a domain name, eg:<a href="/resources/green/mostly/page.htm" ...>Personally, I prefer to
use relative URLs + some absolute (without domain name) ones to shared
- The advantage of being able to test things out 1:1 on a
- server can't compete with the pseudo- protection against
- The only place I would use absolute URLs would be if the
- is hosted somewhere where the webmaster can't do a 301 redirect
and may have trouble with duplicates. I've seen this a lot with sites
hosted on a free account with the ISP; often it will be hosted as
www.isp.com/users/~name/site ..., then perhaps
domain.com/site... and www.domain.com/site... . By using
absolute URLs in that situation, any value passed to one of the wrong
URLs will automatically pass value to the correct URLs as well. If you
have a really good CMS you may be able to change from one to another
and use a staging server without much work. In that case, it probably
doesn't matter which one is chosen. John (*) Regarding the evil
scraper scenario: I think this is overrated and those who have
problems with it usually have other problems to worry about. Also,
most scraper software recognizes absolute links and swaps them out
anyway.
