所谓LTS，就是Long Term Support。这也是Ubuntu在6.06之后的第二个LTS版本。

他们提供了Solaris的系统，并且有C、C++等语言的编译环境。对于想用Unix的人来说，还是不错的。不过没有X。

目前提供几个服务器：

• x4100.unix-center.net
• solaris.unix-center.net

推荐使用Putty使用SSH方式访问。

这样的话，我几乎什么也做不了。我首先想到的是看看日志文件，开始没留意什么，却发现一个log中写着：The directory “/usr/share/…” does not exist。觉得很奇怪，进入/usr，却发现里面什么都没有。/usr里面的内容凭空消失了？觉得问题出在/etc/fstab文件上，查看这个文件，偶然发现，在/dev/hda1的记录上写着：

# Engry for /dev/hda1 :
UUID=a80b3122-dd44-46b7-8fc5-00262a486e32 / reiserfs otail 0 1

注意那个otail了吗？实际上应该是notail。不知道上次在改什么的时候把它给删除了。却发现现在没有vi，没法编辑阿，为了一个字符写个教本也不值当的。

这时我想起了之前介绍的SystemRescueCD， 启动之后，mount /dev/hda1，修改文件，重新启动，搞定！

SystemRescueCd 的 ISO 映像文件的大小为121MB，非常适合刻录为小CD盘片，可从下面下载。

Download SystemRescueCd 0.3.3

Takeaway:
There are numerous common features with every Linux installation. This document lists 10 of the more important ones you should know about.
By Jeffrey G. Thomas

Linux is not Windows, and although there are some similarities, you must realize that there may be a few “new ways of doing things” to learn before you can be comfortable in Linux. Linux is an open-source clone of UNIX, a secure operating system (OS) that predates DOS and Windows and is designed for multiple users. The items in the following list generally apply to any UNIX-based *nix system, such as Linux and the various BSD’s. For the purposes of this article, assume that it’s all Linux.

Here are the 10 things to know
Editor’s note: As part of a recent IT Soapboxblog post I asked Linux users and evangelists in the TechRepublic community to step up to the plate and take a crack at producing some informative articles and downloads on the Linux operating system. This document is just one of the submissions inspired by that challenge. Just click the Linux challenge tag to track other published submissions stemming from this grass roots project.

1. File hierarchy

Unlike some other OS’s that have a file tree for each drive, the Linux file system is one big tree. At the top you have / (Root) and every folder, file, and drive branches off of this Root.

For example, say that you have two hard drives (named a and b), one floppy drive, and one CD-ROM. Let’s say that the first hard drive has two partitions (named a1 and a2). In Windows, it would look like this:

• + hard drive a, partition one (hda1): C Drive
• + hda2: D Drive
• + hdb1: E Drive
• + floppy: A drive
• + CD-ROM: F Drive

In Linux, you have one file system, not the five listed in the Windows example. Each drive is mounted onto the tree and acts just like a folder. The drives could be placed like this:

• + hda1: / (our Root)
• + hda2: /home
• + hdb1: /home/user/music
• + floppy: /mnt/floppy
• + CD-ROM: /mnt/cdrom

Our D Drive and E Drive are attached within our C Drive and there is no need to go to the top to switch drives; the switching happens seamlessly as we move from one folder to another. The same is true with our Floppy and CD-ROM: they are just attached under /mnt as part of the one file system. These drives, in reality, can be attached almost anywhere in a Linux system, depending on how the installation (or user) set up the /etc/fstab file, which tells the computer where things get attached and how to handle them. See Figure A.

Figure A

File structure

2. Modular system

Think about the guts of Windows XP, Win2k, Win98, Win95, and Apple OS X. Each has different components inside and a different graphical look. What if you wanted the Media Player in XP, the File Manager in 98, the stability of Win2k, and the look of OS X all in one unit? In Linux, each aspect of the system is independent, so you can mix and match parts to make your very own Frankenstein OS. You can choose from a variety of programs to run as your firewall, another to play media, yet another to run your File Manager.

Unlike the “tower” OS’s from Microsoft, where everything is interconnected and depends on each aspect of the system, the Linux OS is spread out like a Market: everything works together for the common good, but vendors (independent parts of the OS) can be excluded, and the OS will still function. Don’t want a Media Player or File Manager? Take it out! Your OS will not fall like a broken tower.

This modularity is the reason for so many distributions of Linux (commonly called distros); any person or company can mix and match the programs they find most useful and slap a name on that collection. RedHat, Xandros, SimplyMEPIS, and Suse are all examples of distros.

Some of the larger distros have copycats that use their settings, but change the included programs. This is the Linux way, and the mix-and-match approach gives users more choice in the long run. DistroWatch.com currently lists over 350 distributions of Linux. Many on the list are specialized to serve a specific group of people, but all can be altered to run the same programs.

Because programs are interchangeable, the Graphical User Interface (GUI) is no exception. GUI’s give you the look and feel of a modern OS with the mouse, program icons, menus, etc. Any Linux system (well, one that is running on a real computer, not a phone) can run one of many different GUI’s, just like it can run many web browsers or different email client programs.

Want your system to look like Windows? Use FVWM with the XP theme. Want it to be fast? Try IceWM. Want it to be more “full featured”?; try GNOME or KDE. All of these GUI’s have benefits and drawbacks, but they all present the user with an interface that can be manipulated with a mouse. Although this may result in every Linux screen looking different, all of the GUI’s are still doing the same behind-the-scenes work for you; just use your eyes and often it is not hard at all.

3. Hardware, software, and everything in between

Linux has come a long way in the few short years of its existence. It is less than half the age of Microsoft Windows, and yet it is more powerful, more stable, less resource-hungry, and graphically equal (if not superior) to this costly, buggy OS from Redmond.

One thing that Linux doesn’t yet have going for it is vendor support. If you really like Intuit’s QuickBooks, for example, you cannot natively run it in Linux. There are projects to make Windows programs run in Linux, such as CrossOver Office and Wine, but these work with varying success, depending on the Windows program. Until software companies decide to port their programs to Linux, you will not be able to run them natively.

Not all is lost, however. Open Source software has upward of 15,000 of programs that run natively in Linux. Because these programs are (usually) free of charge, they vary in quality, but the majority of programs are wonderfully written and constantly improved. These programs can import and export non-native file types as well. GNUCash can read those Quickbooks files just fine, and OpenOffice.org can read MSWord *.doc files. If you dislike GNUCash, don’t despair. There are other similar programs also available for free, and more and more software companies are releasing Linux versions of their software.

These same issues apply to hardware. Just as you cannot expect just any piece of hardware off the shelf to work with an Apple computer, the same can be said for Linux machines. Most standard hardware works perfectly; hard drives, RAM, flash drives, motherboards, NIC’s, and digital cameras usually have little trouble under Linux. Newer, cutting edge hardware is a different story. Until hardware vendors choose to support Linux, the drivers needed to work these pieces of hardware must be written by the Linux community for free and in the community members’ spare time.

Therefore, there is a lag behind Windows support since hardware companies often work directly with Microsoft to ensure compatibility, and tend to let Linux volunteers figure out the Linux hardware support on their own. Laptops are notorious for their non-standard hardware; it can be a challenge to map special keys in Linux. The good news here is that vendor support for Linux hardware, like software, is changing for the better as more and more companies see their future in Linux.

Everything in between the hardware and the software in a Linux machine is the kernel. This kernel is what connects the hardware to the software, and an updated kernel is made available via Internet every few weeks; the most current is 2.6.14. If you have hardware that isn’t currently supported, there is a chance that a newer kernel could help you out. Installing this kernel yourself isn’t always easy, however; that’s where Package Managers come into play.

4. Package Managers – Program installation made easy(er)

There are many ways to install programs in Linux, but the easiest is with your distribution’s Package Manager (PM). The PM makes sure that any missing files (called dependencies) are also installed so the program runs correctly. Choosing a distribution often comes down to which type of PM you like, but any Linux software can be installed on different distributions if you can find the corresponding Package.

These PM’s usually have an on-line repository for their programs. Installing an application is as easy as searching through the program repository and clicking Install. Can’t find IceWM or MPlayer in your Package Manager’s list? There is always a way to add a new on-line repository that will have what you are looking for. Some examples of Package Managers include Synaptic (based on dpkg and Apt) for Debian (and derivatives); Yum for RedHat (and derivatives); YaST2 for SuSE (and derivatives); and Emerge for Gentoo.

5. Permissions

Linux is designed to have multiple users, and these users fall into groups. Every user has permissions to read, write, or execute (R/W/X) their own files, and permission to change those permissions. Because Linux is designed for multiple users, each user has their own password and may restrict access to their files. These are called User Permissions.

Each user belongs to one or more groups, and a user can set their file/folder permissions so that others in the group can read but not write the files, or any other combination of R/W/X. These are Group permissions. For example, Joe and Susan are both in Accounting. They can allow the Accounting group access to each other’s files, but they can restrict that access from those in the Sales group.

The Others permissions can allow or deny access to these files for anyone outside the Group. These permissions are for the safety of the overall system, as well as for each user’s data. Most home users are fine to leave the default permissions alone on their files. (See Figure B)

Figure B

Permissions

The Root user (not to be confused with the / Root of a file system), as the Administrator, has rights to all files and is the only user who can alter system-wide settings. The Root user has their own password, which is used for system maintenance. This distinction prevents a regular user from installing harmful spy ware on the system or deleting important files.

6. Home directory

Windows has My Documents, but where do you put files that aren’t documents? Usually on the Windows Desktop! Linux can clutter the desktop too, but each of our users also has a Home directory, usually located at /home/user. Within that Home directory you often have Documents (/home/user/documents), program links, music (/home/user/Music), or whatever we want. We can create files and folders here, and organize or disorganize them as much as we want, just like it was our own personal Home. Depending on how our permissions are set, we can allow or prevent any other user access to these files (except the Root user).

7. Default installation differences

There are a few differences between Linux distributions, such as where some files are kept or what some of the default programs are named. Just knowing that the file system might be a bit different between RedHat and SuSE is a great start. Most users don’t need to know what those differences are, but they should be aware that the internal file systems can be a bit different. When asking for help, make sure to let others know which distribution you are running. If you don’t have troubles in your system or don’t care to set up complex behind-the-scenes operations, don’t worry too much about this.

8. CLI, or “how to run”

From the Start-type menu, the xterm program (also called Console) brings you to a Terminal, which looks a bit like a DOS window, but it actually predates and out-powers DOS. This is the Command Line Interface (CLI), the origin of our favorite OS which is present in every Linux distribution. We won’t get into the finer details, which can fill books, but the CLI a powerful tool often needed to troubleshoot your computer. If you ask for help on the Internet and someone asks you to run lspci, they want you to start xterm, type lspci, hit enter, and then provide the screen’s response.

When you start an xterm, you are your regular user-self with limited powers. To get into Root User mode (see Permissions above) in an xterm, type su [enter], then type the Root password [enter]. Now you have a lot of power so be nice. The Root user can destroy anyone’s data, including the system files needed to run Linux. To leave an xterm or su mode, type exit [enter].

9. Ctrl-alt-escape

Clicking the ctrl-alt-escape key combination changes your mouse into an X, skull-and-crossbones, or some other sinister mouse-cursor. In this mode, clicking on a misbehaving or frozen application will kill it. It is similar to the End Process in Windows Task Manager, but use with care. If you don’t want to kill anything, use the Esc key to back out of kill mode. Clicking on the wrong program (including the desktop GUI) can cause a serious headache.

10. The Internet is your friend!

Many distros have a User’s Forum where questions, answers, and tips are passed around. LinuxQuestions.org is a great site for overall Linux knowledge and help. Remember, before posting questions on any forum, research your questions (both at LQ and on Google) to avoid asking about an issue that may already be solved. Also, check the age of any solution that you find, as old answers may no longer apply to this fast-changing world of Open Source software. When asking a question, be sure to include as much (applicable) information as you can about your system, such as:

• The processor type (Intel or AMD or Apple PPC?)
• Your distribution (SuSE? Debian?)
• The program with which you’re having trouble, and
• Any other relevant information.

Keep an open mind

Linux and open-source software have made leaps and bounds in the last few years, but for users coming from the closed world of Windows, the internal workings of Linux may seem foreign. An open mind and a willingness to share knowledge helps the Linux community grow, and we welcome you to our ranks.

Original posted in:
http://articles.techrepublic.com.com/5102-10877-5960961.html

Or Chinese – Simplefied Edition:
http://wiki.ubuntu.org.cn/10ThingsYouShouldKnowAboutEveryLinuxInstall

Ubuntu的确是汲取了很多Windows中优秀的东西，包括Automatic Update。刚刚装好的系统，右上角的Notification Area就提示了有更新的package要装。更是简单到只要点一下更新的图标，再点一下OK，就可以自动完成更新了。

这里有个不是很爽的东西，Ubuntu（包括Debian）的aptitude，如果他正在工作（下载或安装package），就不能同时开启另一个进程工作。也就是说，如果你在更新系统，那么无论是apt-get还是aptitude，还是Ubuntu的Package Manager都是无法工作。

当然，如果是tarball的package还是可以安装的。

花了将近1个小时下载并安装了全部的更新，其中包括一个linux-kernel-i386-2.6.17-11-generic的package。这里着重说一下，在后面的过程中，正式因为这个新的内核导致了其他的问题。

更新完成重新启动系统，却发现显示分辨率变成了800×600@60Hz了，而且只能从800×600和640×480中进行选择。这其实就是已经存在的nvidia-glx的package和新的内核不兼容。（当初装Debian的时候，就遇到过类似的问题）。

作为暂时的处理方法，先把老内核作为GRUB的默认选项了。至于新内核，我想了一下，-10和-11估计也没什么太大的差距，先凑合用了。

Update 2/23 20:43：翻了翻网上的方法，其实简单的要命。装一下nvidia-glx（如果显卡在“列表”中的最下面部分，就要装nvidia-glx legacy）。然后修改/etc/X11/xorg.conf文件：1、在Module中添加Load “glx”；2、在显卡配置中，将Driver “nv”换成Driver “nvidia”。重启，搞定。

接下来安装en-us环境下的中文支持。Ubuntu在中文支持上和早期的Debian相比，已经有了很大的提高了。想当初早期的Debian Woody默认安装后，是无法显示中文的，所有的中文字符都显示为1个方框，方框的4个角上分别写着1个数字。而现在的Ubuntu Edgy默认就可以显示所有的字符了，虽然看起来并不是那么漂亮。

我打算用文泉驿的字体（当然，用simsun或者Windows Vista中的微软雅黑都可以，只是复制不同的文件而已）。这里有个脚本，可以直接从文泉驿网站上下载nightly build的字体问题，并且自动安装。

[coolcode]
#!/bin/sh
sudo apt-get install build-essential bdftopcf
mkdir temp
cd temp
wget http://wenq.org/daily
tar xzf wqy*
cd wqy*
make
sudo rm *b.pcf
if [ ! -d /usr/share/fonts/wqy-bitmapfont ]
then
sudo mkdir /usr/share/fonts/wqy-bitmapfont
fi
sudo mv *.pcf /usr/share/fonts/wqy-bitmapfont
cd ../..
rm -r temp
sudo fc-cache -fv
if [ -f /etc/fonts/conf.d/30-debconf-no-bitmaps.conf ]
then
sudo unlink /etc/fonts/conf.d/30-debconf-no-bitmaps.conf
fi
[/coolcode]

chmod到755，然后run一下就行了。

仅仅这样做还是不够的，还要修改字体的配置文件/etc/fonts/fonts.conf：
[coolcode]


serif

sudo apt-get install im-switch scim-pinyinim-switch -s scim-pinyin

基本搞定。。

这次之所以选择Ubuntu，首先是因为Ubuntu是基于Debian的一个Open Source的Linux系统，与Debian的“大家投票”的更新形式不同，Ubuntu是有专人（Triage？）决定他采用或者不采用哪些功能或者包。

要安装Ubuntu非常简单，如果是全新系统，而且只安装一个系统，甚至比Windows的安装更加简单。

直接从Ubuntu首页（http://www.ubuntu.com/）上下载最新的版本就行了（到目前为止，Ubuntu的最新版本是Edgy Eft，也就是Version 6.10）。

Ubuntu的版本号比较奇怪，6.10表示它是在2006年10月发布的。上一个版本Dapper Drake是6.06。

如果想得到更多的支持，可以试试看Dapper，因为Dapper是LTS（Long Term Support）的。不过也无所谓，Open Source的最大优点就是有广泛的人使用和支持。

Ubuntu的最近几个版本如下：

• Warty Warthog (4.10)
• Hoary Hedgehog (5.04)
• Breezy Badger (5.10)
• Dapper Drake (6.06)
• Edgy Eft (6.10)

我在下载到Ubuntu光盘镜像后，遇到了个比较棘手的问题。Ubuntu CD是715MB的，而我手里的CDR都是650MB的。没辙，既然先决定要装了，就凑合一下吧。找了张DVD-R给刻了。

Ubuntu CD是个Live CD，意味着即使不安装，也可以直接在光盘上运行。安装更是简单到点击Install图标就可以了。

安装过程还是非常简单的，却在分区中遇到了问题。我因为以前装过Debian，所以就决定用类似的方法手工分区。计划是这样的：

结果有问题，安装都快结束了，提示了一个Install GRUB failed! (fatal error: ‘grub-install’ failed)。想了半天也没感觉有问题啊，然后进入一个Terminal，手工安装GRUB，却提示Can’t find /boot。很怪异。

想来想去，还是暂时放弃了。除了swap，/usr，/home外，都使用一个分区了。结果就变成这样了：

暂时是搞定了。接下来升级系统，发现Edgy居然也有这么多安全更新和包的更新。接下来就是该处理了中文支持了（其实默认情况下，中文已经可以显示了。）

Today, Debian is a troubled distribution. In order to get Debian GNU/Linux 4.0, codenamed etch, out the door by its scheduled date of December 4, 2006, Dunc-Tank.org, a group of Debian developers and fans, was formed.

Historically, Debian has been notoriously late ever since Debian 1.2 was released almost a decade ago in 1997. So, Dunc-Tank was created to accelerate Debian’s development.

If it does die, we’ll miss it — but in Ubuntu and the other Linux distributions that have been based on the best of it, its DNA will live on.

术语解释

在 下文中，我们经常提到 “C-x r” 这类键操作。”C-x r” 其实就是按Ctrl-x，然后按 r。同理 “C-M-@” 就是按 ctrl-alt-@(M表示meta, 在 PC 上就是 Alt 键)，但是其实 @ 是shift-2 (看看你的键盘)。所以 “C-M-@” 实际上要你按 ctrl-alt-shift-2。

但是在配置文件里的键序列中，我们把 “C-x r” 表示为 “\C-xr”, 把 “C-M-@” 表示为 “\C-\M-@”，你自己看看就知道怎么回事了。同理 “Esc a” 别表示成 “\ea”。

这就是 Emacs 里的按键的通常标记方法。EMACS = Esc Meta Alt Ctrl Shift

技巧篇

在自己配置命令行之前，我们先来看看利用缺省的键绑定能够进行的一些巧妙的用法：

第一招：使用以前的命令行参数

你是否经常出现这种情况？你想把 ~/text-browser/ 目录下的3个.tar.gz文件搬到/usr3/software/，于是你输入：

$mv ~/text-browser/*.tar.gz /usr3/software/

我想你一定已经知道，打入 ~/text 之后按 TAB 就可以补全text-browser这个长文件名吧？这是Bash 的基本功能。我废话？好了，就当你知道吧。不过今天我要讲的东西比这个复杂一些。

Go on! 刚刚输入到这里，你突然想起，应该在 /usr3/software/ 下先建立一个目录叫browsers，这样放进去的文件比较好管理。

于是你 Ctrl-u，删掉了这行命令。唉呀，这么长的命令一下就没了。是不是有点可惜？这还不算麻烦。然后你

mkdir /usr3/software/browser
mv ~/text-browser/*.tar.gz /usr3/software/browser

嗯。TAB 是帮了你不少忙。可是你实际上有更好的办法来完成这项工作。好吧，看看 readline 怎样神奇的完成你的任务：我们回到这种情况：

$mv ~/text-browser/*.tar.gz /usr3/software/

你 刚才是按了 Ctrl-u 删除了所有输入的东西。可惜啊！你要是按 M-#(也就是按住 PC 机的 Alt 键，再按 #，实际上就是 Alt-Shift-3)，那么 Bash 就会在这样最开头插入一个 “#”，然后输入这行。这样命令就被作为一行注释载入了历史。这有什么好处？这样你的这行命令里的内容就可以被再次利用。看着：你接着输入：

mkdir ...

等 等，你是不是想输入 /usr3/software/？你不用再敲一遍了！直接按 M-.(Alt 加句号)，看看， /usr3/software/ 是不是出现在命令行上了？M-. 就是调用了 yank-last-arg 函数，把上一条命令的最后一个参数放在命令行上。好了，回车吧！你接着输入：

mv ...

等等，这下是该输入 ~/text-browser/*.tar.gz 了。烦不烦啊？换一种方式吧。请按：”M-1 M-.”(把上一条命令的第一个参数放在命令行上)。这样命令行成为了：

mv /usr3/software/browser

怎么成这样了？看看你的“上一条命令”是什么吧？是……你自己看。所以这个参数不是你想要的。那么继续再按一次 “M-.”。看到了吧？你的命令行已经成为：

mv ~/text-browser/*.tar.gz

好。打一个空格。再按一下 “M-.”。命令行变成了：

mv ~/text-browser/*.tar.gz /usr3/software/browser

这就是你想要的！是不是看起来你还是花了不少工夫？但是想一想，如果你是要执行这样一个命令呢？

mv /data/ftp/pub/TUG/texmf/tex/latex/CJK/GB/GB.cap \
/usr/local/texlive/texmf-local/tex/latex/CJK/GB/

嗯。记住这个有用命令：M-. ， 它的前面可以用 M-0, … 作为数字参数。

第二招：补全命令名，文件名和变量名

你知道 TAB 可以补全命令行上很多东西。可是你遇到这种情况的时候怎么办？

man a-very-very-long-command-name

你 输入了 man a-ver… 之后，按 TAB，什么反应也没有。因为 TAB 执行的是 “按情况补全”(complete)，它看到 man，知道这应该是一个命令，那么它认为： “后面应该是一个文件名参数。” 但是你想要的是命令的名字怎么办？答案：按 “M-!”.再来看：你需要设置 XMODIFIERS=”@im=fcitx”。你输入了

export XM...

按 TAB? 没有反应。为什么呢？因为 TAB 的补全想要一个文件名，而当前目录没有开头是 “XM…” 的文件。那么你怎么补全？答案：”M-$”。其实 readline 的补全方式被 Bash 扩充了很多。看看有多少吧！

"TAB": complete
"\M-!": complete-command
"\M-/": complete-filename
"\M-@": complete-hostname
"\M-~": complete-username
"\M-$": complete-variable

自己试试吧！

第三招：扩展命令行

你 的一个目录里有很多类似的文件，名字叫 T12.txt, T12.log, T23.txt, T23.log, T13.txt, T13.log…… 有后缀 txt 的，也有后缀 log 的。… 你想把其中的某些 T*.txt 都移动到另外一个目录，而T*.log都不动。但是T*.txt 也不是全部都要移动。所以你想把T*.txt 都放在命令行上，然后选择其中一些。你输入：

mv T...

接着按 “M-*”(insert-completions)。结果 T 开头的文件都被放到命令行上了。嗯。这在某些时候是有用的，可是现在它把 T*.log 的文件也放上去了。不行。我们于是继续输入：

mv T*.txt

好了，现在我们可以使用 “C-x*”(先按ctrl-x，然后按*)。结果所有名字T*.txt 的文件都被放到了命令行上面。”C-x*” 执行的函数叫做 glob-expand-word.

配置篇

你是不是觉得那些命令很难记住？不顺手？别怕！它们都是可以改变的，就像Emacs的键绑定那样，可以被任意的改变！

所有使用readline的程序，都使用一个配置文件来决定它的行为和键绑定。这个文件一般是 INPUTRC 环境变量确定的。如果这个环境变量没有值，那么缺省使用 ~/.inputrc。

~/.inputrc 文件很简单，只有4种语句：

1. 注释
2. 变量设置语句(set variable value)
3. 键绑定(“keyseq”:function)
4. 条件语句($if … $endif)

我们先不说其它的，先来看看键绑定吧！

键绑定

1. 绑定语句。你 现在就可以动手设置你喜欢的控制方式。比如，我发现有些时候我需要在命令行上做上 mark(Emacs 术语)，然后把mark 和光标之间的 region(Emacs术语) 删掉，这个操作在 Emacs 里叫做kill-region. 但是我们发现这个函数在 Bash 里缺省是没有绑定的。如果我希望得到跟 Emacs 一样的绑定 C-w 的话，就把这行插入到 ~/.inputrc:

   "\C-w":kill-region

2. 使绑定生效。为了使这个键绑定生效，你需要执行 re-read-init-file 函数。这个函数缺省绑定在了 “C-x C-r”。你修改 ~/.inputrc 之后在 Bash 里输入 “C-x C-r” 就可以使新的配置生效了。
3. 列出可用的函数。不过你怎么知道那些函数可以被绑定呢？readline 的 info 页列出了很多函数，可是你不会每次都去info里查询吧，很麻烦啊。其实你可以使用bash的 bind 命令来得到所有的键绑定:

   $bind -p

   可以显示所有现有的已经绑定和没有绑定的函数。没有被绑定的函数被显示为 “(not bound)”，并被加上了注释。就像这样：

   "\C-g": abort
   "\C-x\C-g": abort
   "\M-\C-g": abort
   "\C-j": accept-line
   "\C-m": accept-line
   # alias-expand-line (not bound)
   # arrow-key-prefix (not bound)
   # backward-byte (not bound)
   "\C-b": backward-char
   "\M-OD": backward-char
   "\C-h": backward-delete-char
   "\C-?": backward-delete-char

   你可以把这个命令的输出作为一个模板，嵌入到 ~/.inputrc 文件。把你喜欢的函数绑定到方便的按键。其实 readline 有三个函数可以让你方便的查询函数，变量和宏的绑定情况，它们是：

   dump-functions
   dump-variables
   dump-macros

   可是它们缺省都没有被绑定到任何按键。你可以为它们分别设置类似 “C-xf”, “C-xv”, “C-xm” 这样容易记忆的绑定。

4. 如果忘了绑定……这样你就可以设置你需要的绑定啦！但是你还是有可能在需要的时候突然记不起哪些键绑定可以补全。这时候你输入：

   $bind -p | grep compl

   得到结果：

   "\C-i": complete
   "\M-\e": complete
   "TAB": complete
   "\M-!": complete-command
   "\M-/": complete-filename
   "\M-@": complete-hostname
   "\M-{": complete-into-braces
   "\M-~": complete-username
   "\M-$": complete-variable
   "\M-\C-i": dynamic-complete-history
   "\M-g": glob-complete-word
   "\M-*": insert-completions
   .......

   这样你记不住一个键的时候就可以方便的查询，这样几次之后，你就会把自己需要的按键都记住了。

配置变量

1. 体验：Bash 的 readline 有一些变量可以控制它的行为。比如:

   bell-style 可以控制出错时是 audible(发出响声)，visible(闪动屏幕)，还是none(什么都不做)；editing-mode 可以控制你是用 Emacs 的输入方式还是用 vi 的；

   completion-query-times 的值控制在补全的个数超过多少N时，bash 提示： “Display all N possibilities? (y or n)”；

   如果我设置 expand-tilde 为 on，当输入“ls ~/doc”，按 TAB 时，命令行会自动变成 “ls /home/wy/doc”.

   如果把 visible-stats 设置为 on，那么列出补全的时候，目录，可执行文件，符号连接，会被分别使用 /, *, @ 来标记，就像 ls -F 的到的结果。

2. 设置：设置的方法极其简单，就在 ~/.inputrc 文件里写入类似语句：

   set visible-stats on

   然后 “C-x C-r” 使设置生效。

3. 怎样知道有哪些设置？可以设置的参数是很多的。使用命令

   $bind -v

   就可以得到所有这些可以设置的变量和它们的值了。