Using UNIX for the JCT TCP/IP Lab


EDIT HISTORY:

18 Cheshvan 5762 (04/11/2001)
Now recommend using ssh instead of telnet.
20 Adar 5761 (15/03/2001)
Updated to reflect the situation of this year.
18 Cheshvan 5760 (28/10/1999)
Updated to reflect the situation of this year.

TABLE OF CONTENTS:

  1. Introduction
  2. Working in the UNIX lab
    1. Logging into a Station in UNIX
    2. Editing Programs in UNIX
    3. Compiling Programs in UNIX
    4. Debugging
    5. Running More Than One Program at a Time in UNIX
    6. Copying Programs from the TCP/IP Lab Web Site in UNIX
    7. Documentation
  3. Working from a Microsoft Windows/NT station

1. INTRODUCTION

For the TCP/IP lab, you will have to know how to do the following on UNIX computers:
  1. log into UNIX computers
  2. edit programs
  3. compile programs
  4. run more than one program at a time
  5. copy programs from the TCP/IP lab web site

I recommend that you do as much of your work as possible on the stations in the UNIX room, because the editor will be easier to use, and because you will be able to use more than 1 window at a time easily. Nevertheless, you can work in one of the Microsoft Windows labs and ssh/telnet to the UNIX lab machines (names: shana200, shana201, etc.).

NOTE: although you may work in pairs, it is absolutely forbidden to give your password to your partner! It is also forbidden to log into your account and leave, letting your partner work in your account.

If you have already given your password to your partner, change it immediately. On UNIX, you can change it via the passwd command. If you have the same password on Lev/Midot, you'll also have to change it there.

You can make files in your account accessible to your partner, via the chmod and/or chgrp commands. Or you can mail each other copies of your source code. See me if you have problems with this.


2. WORKING IN THE UNIX LAB (WYLER BUILDING, ROOM 160)

2.1 Logging into a Station in UNIX

Each station displays a login window. Type in your user name (login name) and password. If you have forgotten it, go to Nachum Danzig (Wyler 232) to get a new UNIX password. Currently, his office hours are Sunday - Thursday 13:30 - 14:30.

2.2 Editing Programs in UNIX

I recommend that you use 'emacs' for editing your files. It has nice support for editing programs, including color-coding different parts of the program (e.g., comments are in red), automatically indenting, etc.
  1. Emacs on-line help
    1. Note the following conventions:
      Convention Meaning Example
      C- control key (CTRL) C-s = control-s
      M- meta key (in the UNIX lab, this is the ALT key). Alternatively, you can use the ESC (escape) key. But it works a little differently:
      1. When using the ALT key, you press it at the same time as the other key.
      2. When using the ESC key, you first press the ESC key, and afterwards the other key.
      M-x = meta-x =
      1. press ALT and x at the same time.
      2. press ESC, and afterwards press x.
      RET RETURN (ENTER) key To go to line 25 of the file:
      M-x goto-line RET 25 RET
      which is:
      1. meta-x,
      2. followed by the string "goto-line",
      3. followed by ENTER,
      4. followed by the string "25",
      5. followed by ENTER
    2. Emacs/Xemacs has its own tutorial. You can probably access it via the Help menu. Alternatively, you can access it via typing C-h (control-h) followed by t.
      NOTE: below are other types of emacs help. These also you can probably access via the Help menu. However, I will list how to access them via the keyboard.
    3. After the tutorial, you can get more information about emacs, the GNU C compiler, and other programs via emacs' 'info' system. To enter info, type C-h i.
    4. In fact, C-h is how you access many help facilities of emacs. There are many others besides the tutorial and info mentioned above. Type C-h ? to see the list of help facilities.
  2. There is also what emacs calls the META key. On the PCs, the ALT key is the META key. M-x means META-x.
  3. To get the colors, click with the mouse on Help --> Options --> Global Font Lock.
    NOTE: in the shiur on 20 Adar 5761, 15/3/2001, I mentioned another method to get colors. Forget it. This method is simpler and better. For example, this method automatically updates the colors; you don't have to type C-l (control-L).
  4. Besides the normal editing commands, here are some commands that will probably be useful for this lab:
    C-x 2
    Split the emacs windows into 2 sub-windows.
    M-x goto-line RET number
    Emacs does not display line numbers. Use this command to jump to a certain line. This is useful, because when the C compiler detetcs an error, it will say on which line it is.
    M-x compile
    If you are viewing/editing C code and run this command from that buffer, this will compile it. The default command it uses is make. If you do not know the UNIX make command, you can change the command line to a gcc command line (see below).

    The compilation, and any messages it generates, will take place in other emacs buffer. If there are any error messages, click on the error message with both the right and left mouse buttons at the same time, and you'll jump to the line of C code where the error was reported.

    NOTE: if there are 3 buttons on the mouse, click on the middle mouse button instead of left and right togther.

2.3 Compiling Programs in UNIX

  1. Compiling with GNU C compiler (gcc)
    1. When your entire program is in one source code file, then compiling is trivial. If the name of the file is myprogram.c, then, to compile:
      gcc -o myprogram myprogram.c

      The -o myprogram option means to store the program in a file called myprogram (without the '.c' at the end).

    2. Now let's say that your program is distributed among two files, main.c and utilities.c, and that you want to call the program 'communicate'. Then you would use the following sequence of commands:
      gcc -c main.c
      gcc -c utilities.c
      gcc -o communicate main.o utilities.o

      The first 2 lines compile source files but does not link them. Thus, the output is not ready to run. Instead, it produces object files that have the same name as the source files, but with '.o' at the end. The third line links the object files & produces a ready to run executable. Note that in the previous example with myprogram.c, the compiling and linking were done via one command.

    3. You can compile a gcc program from within emacs. Go the window that contains the program you want to compile. Then I think you type M-x compile.
    4. You can get more information about gcc via emacs info mode, and via man gcc.
  2. If you want to write in C++, then use the command g++. This calls the GNU C compiler, but in a way suitable for C++.
  3. When you hand in an assignment, you must indicate the exact command(s) used to compile it.
  4. See above for compiling the program from within emacs.
  5. You may use the UNIX make command if you want. But we won't have time in this lab to teach you how to use it.

2.4 Debugging

If you compile with gcc, you can use the 'gdb' command to debug. Read about this command in emacs' 'info' mode.

There should also be a debugger command called 'ddd'. This has a graphical interface.

Of course, you can always use 'printf' or 'fprintf' to display debug and error messages.

2.5 Running More Than One Program at a Time in UNIX

This is very simple when sitting at a UNIX station: simply open 2 or more windows, and run a program in each of them.

2.6 Copying Programs from the TCP/IP Lab Web Site in UNIX

Simple. View the program via Netscape. Then save the file in text format.

2.7 Documentation

There are several ways to get documentation about UNIX commands and functions (including socket functions):

Gnome Help Browser
If this does not start automatically, then look at the panel at the bottom of the screen. There should be an icon with a question mark on it. Click that. From this browser you can access all (or almost all) of the other types of documents listed below.
man pages
This is the classic UNIX documentation. The 'manual' is divided into several sections. For example, section 1 are normal user commands. Section 2 are system calls (including the socket interface). Section 3 are other standard routines, etc.

Here are different ways to view the man pages:

  1. From the unix command line, type the 'man' command. Here are examples:
    • man xterm -- learn about the xterm (X windows terminal emulation) program.
    • man 2 socket -- learn about the 'socket' function (which is in section 2 of the manual).
    • man 4 socket -- learn about the socket programming interface (in section 4 of the manual).
    • man -k socket or apropos socket-- get a list of all man pages dealing with sockets (actually, get a list of all man pages whose summary lines contain the word 'socket').
  2. Type xman to get an X windows interface to the 'man' command. From there, you can click on 'Getting Started' to learn how to use xman, or 'Manual Page' to see the man pages. The sections of the manual are the same as for the 'man' command above.
  3. From within emacs, type
    M-x RET man RET man_page.
    If you want specify the manual section, do it like this:
    • socket(2)
    • socket(4)
Emacs info
See the section on emacs info above.

3. WORKING FROM A MICROSOFT WINDOWS/NT STATION

Again, I strongly urge you to work in the UNIX lab, or at least from a UNIX station. But there might be times when that is not practical.

3.1 Logging into a Station on a PC

I assume that you know how to log into a PC. After that, however, you will have to ssh/telnet to one of the UNIX lab stations (shana200, shana201, ..., shana220).

I strongly suggest that you use 'ssh' (Secure SHell) instead of 'telnet'. Ssh has the following advantages over telnet:

To use ssh, run the 'putty' command. There might be another SSH program available on our MS Windows stations as well.

If you cannot use ssh, I suggest that you use the EWAN telnet program instead of the standard Microsoft telnet program.

Whether you use ssh or telnet, you enter the same user name (login name) and password as you do when sitting at a UNIX station.

3.2 Editing Programs on a PC

You'll still run emacs. However, you should run it with the -nw option. This option stands for 'no windowing'; it will prevent emacs from trying to start up an X window (X is the windowing system on UNIX). Instead, it will take over your telnet window.

Alternatively, if you know the vi editor, you might want to use that.

There might not be a META key when working like this. In this case, use the ESCAPE key. See the emacs tutorial for how to use the escape key.

3.2 Compiling Programs on a PC

Same as compiling programs for UNIX.

3.4 Running More Than One Program at a Time on a PC

There are 2 ways:
  1. You can run the ssh/telnet program more than once.
  2. Use emacs. Emacs can split the screen of an ordinary text terminal or of a telnet window. Type C-x 2 to split the screen into 2. Type C-x o to move the cursor to the 'other' window.

    You can run normal UNIX commands from within emacs. Type M-x shell to start a UNIX shell. This shell buffer will be called '*shell*'. To create another shell buffer, you must rename the first shell buffer via the M-x rename-buffer command and then typing M-x shell to start a second shell.

3.5 Copying Programs from the TCP/IP Lab Web Site on a PC

Viewing the program via Netscape or Internet Explorer and then saving it will save it on the PC (either on the local disk or in your LEV home directory). But you need the file in your UNIX directory.

You cannot run Netscape from UNIX and have it display on your PC screen. But, from UNIX you can run 'lynx'. THis is a web browser that works in text mode, so you can use if from within the telnet window. After you view the file via lynx, save it; the file will be in your UNIX home directory.


This document maintained by Haim Roman, roman@mail.jct.ac.il