Linux 101 (for Dummies)

(Actually this is notes to self so I do not need to look this up from the web every time I need it!)

Teletype and Other Arcane Stuff

Traditionally Unix based system are based on command line interface used over a terminal connection. Even if it all runs in a single PC. This is very powerfull but for those born into GUI enviroments it can be a bit intimidating and it has some features that are based on ancient computer history. Things that nobody bothers to explain.

Incidentally, this is part of University 101 - Computer Usage, worth one credit. If you know 1/10th of what's on this page, you'll pass. I did.

Back in the good old days, when I was in my early teens, computers were used via a terminal called Teletype. It was basically an electric typewriter where you punched keys to enter commands into the computer and the computer hammered (literally) it's output to roll paper. What is still significant from this old interaction model is that computer output was and still is static. So if you list the contents of a directory , and then delete a file, the listing stays the same, unless you take another listing. Not like a modern GUI window which tries to be WYSIWYG. (Ok, Windows does a bad job at this but just keep pressing that F5/refresh button...)

An other rudiment is that because output and input at 110 bps was slow and NOISY (try shouting RAT-TAT-TAT-TAT to get a feel of it) you wanted to minimize the number of characters used, hence most Unix descendants have very terse command names and output formats.(Yes, 110 bps is 110 bits/second or about 10 characters/second or to give you some perspective my WLAN is 500 000 times faster! Those were the days, and yet its only a quarter of a centry ago.)

So the main interaction model in Linux is a terminal session using a 'glass teletype'. You type in commands and watch the outpour of characters on the terminal. A session begins when you log into a Linux system and ends when you log off. It is usually possible to open multiple sessions by launching the terminal window multiple times. This can be very usufull as it is like working on multiple computers at the same time.

Of Files and Directories

All large systems, which today means any system, have thousands of files. On this laptop that I'm writing this on I have about 100.000 files. So obviously you want to organize them some how so that you can find anything at all. So the files are organized into directories that can contain further directories and/or files. This is called hierarchical file system. (Pity that the QDOS that Bill bought back in the old days and that later became known as MS-DOS did not have a file hierarchy, not to mention long file names. If he had bought something deacent, like OS-9, boy would we all have been saved from a lot of headache.)

Unlike Windows, Linux uses no drive letters, such as 'C:' to refer to disks. In Linux everything is a directory. The top most directory is equivalent to the drive letter. In Cygwin, which pretends to be a Linux but runs on Windows, the windows 'C:' disk is refered to as directory '/cygdrive/c'. And the root of Cygwin directory tree looking from Windows side is (usually) 'C:/cygwin', under which you find 'home', 'bin', 'etc' ...

File and directory names can contain all kinds of letters but it is best to use names that only use english lowercase letters and digits plus maybe dashes and underlines. Especially you should avoid names with spaces (' ') in them. They will make your life miserable. Names that begin with period '.' are not usually visible in directory listings which is why the many operating system tools store configuration data into files beginning with '.'. Or actually it is the other way round, because tools store info into those files they are hidden.

Names are case sensitive in Linux whereas in Windows they are not. This can be very confusing if you are testing something, say a web page, in Windows and it works but does not work when you upload it to the server which is most likely a Linux box.

Each session has current directory assosiated with it. What this means in Finglish is that if you do not specify (in a command) otherwise it is assumed that you are refering to the current directory. By having multiple sessions it is easy to work in multiple directories because in effect every terminal window has its own current directory.

Here are the most common commands that are used to work with (current) directories.

 pwd                Show the directory you are currently in
 cd dirname         Change current directory to 'dirname' directory 
 cd ..              Move to (make current) the parent of your current directory
 cd ~               Move to your home directory
 mkdir dirname      Make a new directory named 'dirname'
 rmdir dirname      Remove a directory named 'dirname'  Note:  All files must first be deleted

Paths

There are relative and absolute path names. A relative pathname begins with a letter and it refers to a file or directory that is in the current directory or a subdirectory of current directory. An absolute path name begins with a slash and specifies the full 'path' from the highest level of the file system to the file/directory in question.

The current directory can be referenced as '.' and the parent directory of the current directory as './..' and so on.

Every user, based on their username, has a special directory called home directory. Immediatelly after login current director is set to your home directory. The home directory can be referenced as '~' in a path.

The asterisk ('*') can be used as a wild card in most situations when specifying files on a command line.

There are conventions in Linux world on how to organize the file hierarchy. There is nothing to enforce these conventions but most system are organized along these lines.

Here are some often used directories:

 /bin               for binary (executables),
 /dev               for device files,
 /etc               for admin and personal information, 
 /tmp               for temporary files, and
 /home              for home directories of individual users 

Working with Files

 ls                  List the contents of the current directory
 ls -l               Give a long-form listing, with lots of information about each file and directory
 ls -a               Do a normal ls, but include all files whose names begin with a period (hidden files)
 ls -F               Do a normal ls, but mark all executable files with a *, and all directories with a /
 ls -laF             Do all of the above
 ls -R1              List all files in current directory and all sub directories, each file on its own line
 
 mv oldname newname  Move (or, more precisely, rename) the file oldname to the file newname.
 mv file dirname     Move a file into directory named dirname.

 cp filename newname Copy the file filename to the file newname.
 cp filename dirname Copy a file into a directory.

 rm file             Remove (that is, delete) a file.  There's no "undelete" command, so be careful.

 cat file            "Concatenate" the file--that is, print its entire contents to the screen, all at once.  
                     For big files, use more file.

Getting Help

ls --help

man ls

Commands, Scripts and Shells

In DOS/Windows you can store commands into batch files that end with the name extension '.BAT' which allows you to execute those commands by typing the file name. In effect creating more commands out of existing commands. This is often called scripting and the the files that contain commands are called , you guessed it, scripts.

In Linux any text file is a potential script. When you type a file name the shell checks if the file is marked executable and if it is it will execute the commands in that file. A nice example of this is the way GNU tools are configured. You usually configure them by going to the directory that contains the source code and type

configure something

To make a file executable (for everybody) you use the chmod command, like this:

chmod a+x filename

To see if a file is executable and other stuff use:

ls -l filename

Pipes and more (or less)

ls >mydirlistfile.txt

ls -R | less

Using Grep as a Filter

Simply (very simply) put grep grep reads the input and passes through lines that match the criteria you give to the grep command.

Anyway, here are some usefull grep's :

grep ".java"     Pass through every line that contains the string ".java"
grep ".java\|.c"     Pass through every line that contains either the string ".java" or ".c"
grep -v ".class"     Pass through every line that does NOT contain either the string ".java"

Following is a way to list all '*.java' files and the directories in which they are starting recursively from current directory:

ls -R1 | grep -e ".java\|/"

Finding (in) files

grep -n void example.h

grep -n 'int x' *.h

grep -Rn 'func' *.h

grep -Rn 'b*lean' filename

Counter intuitively and unfortunatelly the following DOES NOT work:

grep -Rn void *.h

If you want search all files of certain type in current directory and sub directories you need to use:

find . -name "*.c" -exec grep -l -n 'this' {} \;

Finds 'this' in any '*.c' file.

Following gibberish executes wc< (word count) command on each '*.java' file in current directory and any subdirectory and stores the results to a text file 'wc.txt'. Handly for collecting statistics of you lates software project.

find . -name "*.java" -exec wc {} \; >wc.txt 

Unpacking Files

tar xfj collection1.bz2     Uncompresses and then extract all the files from the arcieve 'collection1'
tar xfz collection1.tar.gz  Uncompresses and then extract all the files from the arcieve 'collection2'

Most Linux/Free software is delivered in source code format.

makeing Programs

./configure
make     
make install  

Compiling Software

#include <stdio.h>

int
main (void)
{
  printf ("Hello, world!\n");
  return 0;
}

cc hello.c -o hello

and executed (run) with:

./hello

thats all folks,
cheers Kusti

Miscallenious

find DIRECTORY -type f -exec md5sum "{}" \; | sort >/tmp/index

find . -name "*.java" -exec grep -v '^[[:space:]]*$' $ {} \; | wc

Current Directory is NOT in the Patch