Tips for using Unix

X windows

• xterm keyboard shortcuts: I have a file ~raphael/.Xdefaults in which I override some keys on the keyboard to give them new meaning. (Look for the line starting xterm*vt100*translations.) For instance, I find that I often want to repeat the previous shell command, but this time pipe the result through the less program. In csh, the command I need is !! |& less. I assign that expansion to the F11 key, which has no particular meaning to xterm.

  To get these settings to take effect, I run xrdb ~raphael/.Xdefaults from my X startup script.

Mail processing

• I use mutt as my MUA (mail user agent). I have some nice shortcuts for responding to mail and including the correspondent's name at the top.
  1. First, in my .muttrc file, I have these lines:

     		set editor="vim -X"
     		set attribution="On %d, %n wrote:\n>"
     		macro pager \CA 'r^m^m^m^m' 
     		macro index \CA 'r^m^m^m^m' 

     Now when I want to respond to mail, I just type <control-A>. Whether I am looking at the list of messages ("index" mode) or a particular message ("pager" mode), the macro expands to the r command, which means "reply", followed by answers to the questions that mutt will ask about whether I want a particular subject and other questions.
  2. Once I am in the vim editor, I see the message to which I am responding, with > at the start of each line. Now I use another macro, which I define in my ~/.vimrc file:

     		map , 0!+extractSender.pl^Mj 

     where the ^M is an embedded carriage return. This macro gives me a command invoked by the comma character that replaces the first line of the message with the result of my extractSender.pl program. (If you like it, you can find it in ~raphael/cs/bin/extractSender.pl.) This program knows how to extract names from the "from" line of a message based on the attribution setting in the .muttrc file, as shown above. So if the attribution line says:

     		On Fri, Sep 01, 2006 at 08:25:17AM -0400, Alfred E. Neuman wrote: 

     The "," command replaces that line with

     		Alfred, 

     and I am ready to continue editing my response.
  3. Sometimes I want to use a nickname for somebody, or extractSender.pl can't figure out a good name for somebody. In that case, put a line in ~/.namelist, such as

     		Alfred E. Neuman => Big Al 

     and extractSender.pl will use what you have specified instead of its usual behavior.
• mutt has the concept of mail folders. These folders are usually found in your directory ~/Mail, although you can configure that with set folder="wherever" in your ~/.muttrc file. mutt names these folders with a preceding = symbol, so it calls ~/Mail/foo by the shorthand =foo. You can save a message to a folder by s =foo. You can switch attention to a folder by c =foo. I find it helpful to have a symbolic link ~/Mail/regular pointing to /var/spool/mail/$USER. That way I can easily switch between special folders and my regular mail.
• Mail attachments indicate what their type is, and mutt uses that type information to decide what program to call to show you the attachment. You specify these programs in a file called ~/.mailcap. Here are some lines I find helpful.

  	application/msexcel;gnumeric %s
  	application/powerpoint;soffice %s
  	text/html; firefox %s;
  	application/ms-tnef; tnef -w %s
  	application/pdf;xpdf %s 

  These lines allow me to view spreadsheets with gnumeric, presentations with OpenOffice, HTML with firefox, winmail.dat attachments (MicroSoft) with tnef, and PDF files with xpdf. If I want to see an HTML attachment without directing it to firefox, I give mutt this command:

  	| lynxnow 

  The lynxnow script, which I put in ~/bin, just has this line, which is too hard to remember and type each time:

  	lynx -force_html -stdin 

• If you use Thunderbird as your MUA, you can download a plugin called LookOut that lets you view winmail.dat (TNEF) attachments.
• Incoming mail passes through several filters before you see it: It must survive greylisting (spam sent by spambots does not survive), scanning for viruses and trojans, and evaluation for spam. You have control over the very last filter, which is called procmail. You may specify rules for that filter in a file called ~/.procmailrc. This file contains rules written in a cryptic style. Here is an example.

  :0H # If it has a very high spam score, toss it
  * SpamScore: \$\$\$\$\$
  /dev/null

  The first line says, “Here is a rule (:) that looks at the message header (H) and is helpfully commented to refer to spam scores (# ...)”. The next line says, “Here is a condition (*) that matches the pattern SpamScore \$\$\$\$\$, that is, the word SpamScore with at least five dollar signs.” The third line says, “If that condition matches, put the mail in /dev/null, that is, discard it and don't consider any more rules.”

  Here is a pair of rules I have found very useful:

  :0hw
  EXTERNAL=| perl /homes/raphael/bin/checkMail.pl
  
  :0
  * EXTERNAL ?? bad
  /dev/null

  The first rule sets an internal variable EXTERNAL to the result of passing the header to my checkMail program. This program either prints "good" or "bad". The second rule sees if checkMail said "bad", in which case it discards the mail. I update checkMail periodically to notice spam that is not filtered out by the department's spam-scoring software.

  You can use procmail to sort your incoming mail into mailboxes based on information in the header. I don't find that useful, but many people like this idea.

• If you want to set up a mail reader such on an Android or Apple device, or you want to read mail with Thunderbird, Outlook, or Apple mail, see this site built by Matthew Ruffner.

Appointment calendar

• Years ago, a student of mine in Wisconsin (Al Michael) built the app appointment-calendar program. I enhanced it in various ways. I find it very useful. Among its nice features:
  1. It is purely text-based, so it is easy to use across the network. I keep my appointments on the CS computers, but they are easily visible when I use ssh to connect to those computers.
  2. You can run app in the background to give you warnings about events that are about to happen. For instance, app -p -t5 & warns me 5 minutes before any event. I start a few such background processes (for 5, 10, and 30 minute warnings) in the same window in which I keep an instance of the mail reader mutt running.
  3. When an appointment becomes obsolete, a record is stored so you can retrieve it if you need to.
  4. You can easily set up appointments that only apply on the third Wednesday of every month except February, if you want.

Diagrams and figures

• I used to use the gremlin program along with troff to draw figures and embed them in documents, but neither has been supported for years. Now I use xfig. It is an interactive program that lets you draw boxes, text, and arrows, with different color backgrounds, at different levels (so that objects can avoid obscuring other objects). It saves files with its own format, but it can also export in PostScript, GIF, PNG, and other formats.
• It is easy to embed a picture in LaTeX. If the picture is in embedded PostScript, use these commands:

  \usepackage{epsf}
  	...
  \epsfbox{fileName.eps}

  If you are using pdflatex, you will also need to

  \usepackage{epstopdf}

  .
• If your picture is the wrong size, you can scale it with

  \scalebox{number}{\epsfbox{fileName.eps}}

  where number is smaller than 1.0 for reducing or greater than 1.0 for enlarging. Alternatively, you can specify exact size with

  \resizebox{hlength}{vlength}{\epsfbox{fileName.eps}}

  Either length can be simply ! to use the same scale factor for both dimensions.
• In many cases, you can export an xfig picture as LaTeX and simply include the resulting file in your LaTeX document. If your picture has text with LaTeX commands, that is, it uses special characters like these: $ { } \, you should build such text in xfig with the "special" flag turned on. That's one of the Text options on the bottom line, right before Text Size.
• The graphviz suite of programs lets you draw both directed and undirected graphs with high-quality automatic layout. The LaTeX package dot2texi lets you embed these graphs as pictures in a document and get the full power of LaTeX, including mathematical notation. Output can also be in many formats, such as PNG and embedded PostScript.
• The convert program can convert pictures from one format (such as GIF) to another (such as PNG), modifying parameters (like size and cropping) as it goes.
• The shareware xv program is very useful for grabbing screen shots of windows or regions of the screen, manipulating those images (size, color map, smoothing), and then saving in various image and document formats. It is also a good picture-viewing program.
• The eog program is another good picture-viewing program.
• For fancy picture manipulation, you need to use gimp, which is full-featured but hard to learn.

Text editing

• There is seldom any reason to use Microsoft (MS) Office™ products. Here are some useful replacements.
  1. Text editor: Instead of MS Office™, use OpenOffice (invoke by soffice) or AbiWord (invoke by abiword). Both are able to read text files created by MS Office as well as a host of other proprietary programs. They have their own output forms but can also create files that are acceptable to MS Office. If you use one of them to read MS Office documents, most of the formatting is exactly right, but there are likely to be minor discrepancies, particularly if the MS Office document uses strange fonts. AbiWord starts much faster than OpenOffice, but it is less fully featured and less polished.
  2. Presentations: Instead of MS PowerPoint™, use OpenOffice. It can read PowerPoint™ documents, but bullets look different.
  3. Spreadsheets: Instead of MS Excel™, use either OpenOffice or gnumeric. The latter is faster to start up.
  4. Databases: Instead of MS Access™, use OpenOffice.
• You can place non-Latin characters, such as "ę" and "א", in documents when you type them with the vim text editor. Several settings must conspire to make these characters work.
  1. You must pick an encoding. I strongly suggest you use Unicode, in particular, the UTF-8 encoding. In your ~/.vimrc file, you want to set

     :set encoding=utf-8 fileencodings=utf-8,latin1

     to tell the editor that it should use UTF-8 as its display encoding and that it should assume that files it reads in are in UTF-8 unless that fails, in which case it should fall back to Latin encoding.
  2. You must enter the characters using a technique known to vim. There are several alternatives.
     • The easiest way is to use a digraph, which is a two-letter abbreviation. You can see the list of digraphs by giving vim the command

       :digraphs

       To enter a digraph, type control-K followed by the two letters. For instance, to get "ę", type "<control-K>e;".
     • If you know the Unicode value of a character, you can enter it even if it has no digraph by typing "<control-V>uXXXX" where XXXX represents four hex digits, such as 05d0 for the א character.
     • If you often need unusual characters, consider using keymaps. You can see an example in my .vim/keymap directory. This effort is only worthwhile if you use non-Latin alphabets extensively.
  3. You must tell your terminal to display Unicode characters. For xterm, put

     xterm*utf8: 1

     in your ~/.Xdefaults file, as described elsewhere in this file. For gnome-terminal, adjust Terminal->Set Character Encoding.
  4. You must select a terminal font that contains Unicode characters. For xterm, I put the following in my ~/.Xdefaults file:

     	xterm*font4: -misc-fixed-medium-r-normal--20-200-75-75-c-100-iso10646-1
     	xterm*Font5: -misc-fixed-bold-r-normal--20-0-100-100-c-0-iso10646-1
     	xterm*Font6: -misc-fixed-bold-r-normal--24-0-100-100-c-0-iso10646-1

     and I generally use font 4. (Click on the right mouse button in the xterm window and select "Medium" font.) For gnome-terminal, edit your default profile to set the font to one that understands Unicode, such as Bitstream Vera Sans Mono.
  5. You can't just send non-Latin text to the printer and expect it to print correctly. Instead, you need to use fancier tools like TeX.
  6. You can place non-Latin Unicode characters in HTML web documents, but your reading public must manually tell its browser to treat such documents as encoded in UTF-8. If you name your file whatever.utf8.html, then our web server automatically tells the browser to use UTF-8, saving your reading public some grief. Another technique is to spell out all non-Latin characters. To embed, say א (that's Unicode 05d0) in a document, use "&#x05d0;". Vim can tell you the Unicode of any character; just place the cursor on the character and type the command ga ("g" stands for "gobs of extra commands that don't fit in the limited number of keys", and "a" stands for "about the encoding".)
• Vim has many useful features. You can place all the commands below for establishing the features in your ~/.vimrc file.
  1. Checking spelling.
     1. Establish by :set spell spelllang=en. Vim now highlights every word it does not recognize.
     2. You can tell Vim to permanently recognize a word with the command zg, or temporarily (for this session) with the command zG.
     3. You can switch to another language. For Polish, for instance, run the command :set spelllang=pl. If Vim doesn't yet have a dictionary for that language, it will help you download a dictionary.
  2. Formatting text so that lines are about the same length.
     1. Establish your preferred length with :set textwidth=80 (say).
     2. You can have Vim automatically move to the next line while you enter text that exceeds the line length with :set wrapmargin=1. Wrapping now starts 1 character from the end of the line.
     3. The gq<motion> command formats text between the current location and the result of the motion. For instance, gq} formats to the end of the current paragraph. Formatting means putting as many words as possible on each line, but not exceeding the text width.
  3. Editing files that have syntax. Vim can recognize files whose content follows syntax rules, such as programming languages and TeX and HTML files, and display special content in color. Reserved words are in yellow, comments are in blue, constants are in red, errors have red background, and so forth. (Of course, you can customize the colors.)
     1. Establish syntax coloring by :syntax on sync fromstart.
     2. Establish file-type recognition by :filetype on.
     3. You can move the cursor from one token to its matching token with the % command. For instance, if you are editing an HTML file, and the cursor is on <table>, this command moves the cursor to the matching </table>.
• There are times when you need to convert tab characters to spaces or spaces to tab characters in a file.
  1. Use expand -4 < filename to convert all tabs to 4 spaces. The 4 represents the tab-stop setting; you can use any reasonable number.
  2. Use unexpand -4 < filename to convert spaces to tabs. Again, the 4 represents the tab-stop setting.
  3. I often call expand and unexpand from Vim, using a command like this to expand the tab characters in a paragraph: !}expand -4
• You can easily check spelling in a text or PDF file. I use the following csh aliases:

  alias spelltext 'ispell -l < \!\!$ | sort -u | less'

  alias spellpdf 'pdftotext \!\!$ - | ispell -l | sort -u | less'

  Given those aliases, you can find all the questionably spelled words in a PDF document like this:

  spellpdf whatever.pdf

• If you want to combine PDF files or extract pages from PDF files, try using pdftk. For example, to extract pages 2-5 from foo.pdf into bar.pdf, do this: pdftk foo.pdf cat 2-5 output bar.pdf. The pdftk program can merge and split documents, rotate pages, decrypt and encrypt (with password), generate a stencil from PDF forms and use a filled-in stencil to populate those forms, apply a watermark or a foreground stamp, and uncompress and recompress a document.
• You can add simple text annotations to a PDF file by using flpsed.
• You can add more complex annotations to a PDF file by using xournal: highlighting, page resizing, text in Unicode, deleting pages, adding pages.

Version control

• If you have a single file that you occasionally change, but you want to store old copies, you might use a simple version-control application. I do that for letters of recommendation (usually only the address changes) and programs that I am fiddling with. A version-control application can show you what versions you have, when you created them, and what comment you associated with them when you created them. It also lets you retrieve any of the versions. I recommend rcs for simple version-control tasks.
  1. To create a new repository for file foo: ci -r2 foo The ci command means "check in". The -r2 parameter says "make this revision 2"; I do that so you get prompted twice: once to describe the file, and a second time to describe the revision. The result is a file called foo,v.
  2. I usually unlock the repository, because I don't need to lock working copies: rcs -U foo.
  3. To check out the most recent version: co foo
  4. To check in a new version: ci -u foo The -u means "keep a regular copy of foo in addition to the checked-in copy."
  5. To see what versions there are: rlog foo
  6. To check out version 4.3 of foo: co -r 4.3 foo
• If a collection of files together make up a project or multiple users need to coordinate their changes to a project, you need a more complex version-control application. I recommend subversion. Once a repository for the project is set up on some server machine, each user extracts a private copy. In all cases below, I use foo to refer either to a single file or to a directory, which could be the top-level directory of the private copy.
  1. Merge other users' changes to foo into my private copy: svn update foo
  2. Update the repository for foo based on changes I have made in my private copy: svn ci foo
  3. Add foo to the repository: svn add foo
  4. Remove foo both from my private copy and from the repository: svn rm foo
  5. See the status of my copy of foo with respect to the repository: svn status -u foo (The -u indicates that the program should not only report local changes, but it should query the repository for changes that others have made.)

TeX processing

• I have built a make script to help me process TeX (or more specifically, LaTeX) files. You can find it at /homes/raphael/cs/papers/Makefile.tex on our machines. Here is a sample Makefile that uses it:

  GOAL = slides
  SOURCES = insert.tex
  EPS = fig1.eps fig2.eps
  PUBLISHAS = conf.2006.02
  
  include /homes/raphael/cs/papers/Makefile.tex

  The GOAL definition specifies the main TeX source file, in this case, slides.tex. If you have subsidiary TeX source files, list them in the SOURCES specification. If you have encapsulated PostScript figures, list them in the EPS specification. If you want to place the processed document in PDF form on your web page at /homes/yourUserName/HTML/private/foo.pdf , specify foo in the PUBLISHAS line. You may omit any of these specifications except GOAL.

  Once you have built such a Makefile, you can now try the following commands:

  make dvi	calls latex to create GOAL.dvi, converting .xfig files to .eps if needed.
  make seeDVI	builds GOAL.dvi and then runs xdvi to view it
  make ps	builds GOAL.ps
  make seePS	makes GOAL.ps and then runs gv to view it
  make pdf	builds GOAL.pdf
  make seePDF	makes GOAL.pdf and then runs xpdf to view it
  make publish	makes GOAL.pdf and then copies it to /homes/USER/HTML/private/
  make spell	displays a list of possibly misspelled words (using ispell and less)

• If your document needs an index, here's how to get one.
  1. Before \begin{document}, place \makeindex.
  2. At every point in your document that you want your index to refer to, place \index{major!minor}, where major is the principal index heading, and minor is the subordinate heading. You can leave out !minor if it doesn't apply.
  3. At the end, where you want to place the index, place \section{index}\input{filename.ind}, where filename is the name of your source file.
  4. Run makeindex filename each time before you run latex.
• Sometimes you need to create a PDF with all fonts embedded. Starting with a PostScript file input.ps, you can create output.pdf this way:

  ps2pdf14 -DPDFSETTINGS=/prepress -dEmbedAllFonts=true input.ps output.pdf

  You can see what fonts a PDF file uses with

  pdffonts filename.pdf

HTML authoring

• I find that the vim editor is very good for editing raw HTML; it is able to match opening and closing tags, for instance, and show you by color changes when you have made a mistake such as failing to close a string. It does get confused, though, when you place javascript code in the middle of a page.
• Most browsers are very forgiving of HTML errors. Nonetheless, it is good to know that your HTML is free of errors. I suggest you use the HTML validator service for that purpose.
  1. You can direct your browser to http://validator.w3.org/ and enter the URL of the page you wish to validate.
  2. You can make a bookmark in your browser with location set to javascript:window.open('http://validator.w3.org/check?uri='+escape(window.location));void%200 and name "Validate HTML". Going to that bookmark has the effect of running the HTML validator on your current page.
• You most likely want to place the following at the very start of your HTML document to inform the browser (and the validator) what standard you are trying to meet: <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">

File searching, saving, and deletion

• The find program is extremely useful for searching through a directory of files.
  1. To list all regular files (not directories) below the current directory, try this:

     find . -type f -print

  2. To just list the directories, try this:

     find . -type d -print

  3. To search all Java files for the string myVar, try this:

     find . -follow -name '*.java' -type f -exec grep -l 'myVar' {} \;

  4. I use that sort of search so often that I have built a csh alias for it:

     alias rgrep "echo find . -follow \\
     		-name \'\!\!:2\' -type f -exec grep -l \'\!\!:1\' {} '\;' "

     I use it this way:

     rgrep myVar '*.java'

     I then copy and paste the resulting find command.
• Every so often I run /homes/raphael/bin/make_myfind.sh to build a file ~/.filefind.data listing all of my files. I can then search for all my files whose names contain a given substring by

  /homes/raphael/bin/local.i686/afind "the string" ~/.filefind.data

• The locate command (in Ubuntu Linux) queries a database of files local to your computer that is updated regularly.
• If you have lots of spare local disk, you might want to create (or have created for you) /localdisk/home/saved.YOURLOGNAME. Using crontab -e, add this line to your list of commands run according to a schedule:

  30 3   *   *   *     /homes/raphael/cs/public/localdump

  That command will run localdump every morning at 3:30am to save a copy of your incoming mail and all your recently created or modified files in /localdisk/home/saved.YOURLOGNAME.
• If you want to make sure that a file containing sensitive data is truly gone, you might want to run shred on it. This program repeatedly writes random bytes across the file and then (optionally) removes the file. The reason multiple writes are necessary is that a single write changes the magnetization of the track on the disk, but doesn't necessarily change magnetization a trifle to one side, which can still be read by expensive equipment. Several writes will generally completely obscure the former content. The shred program also works on hard drives directly (such as /dev/sda1) if you have permission to write to them. Warning: shred does not remove backed-up copies of the file, and it doesn't work on RAID filesystems.

Accessing the web

1. Usually you access the web using a browser. You are likely to be happy with firefox. If firefox crashes, it sometimes leaves a lock file that prevents you from starting it again. You can remove that lock this way:

   rm -f ~/.mozilla/firefox/*/{lock,.parentlock}

2. You might prefer opera or konqueror. For very fast start up but limited features (no JavaScript, for instance), I can recommend dillo. Not all these browsers are installed on our machines, though.
3. If you find yourself in an environment where you need a text-only browser, consider lynx or www-browser.
4. If all you want to do is fetch a file given a URL (whether it is http or ftp), you might be happy with wget.
5. The ncftp program is an enhanced ftp client, knowing how to log in anonymously and remember places you have visited. Such a client lets you navigate the file system of a remote site, if the site allows such ftp access.

Shell commands

1. Unix provides many different shells to choose among. I prefer tcsh; others like bash. Both provide a way to re-submit commands from your recent history and to edit those commands. Both can read start-up files to establish your favorite settings. For tcsh, the start-up file is ~/.tcshrc; for bash, it's ~/.bashrc.
2. tcsh lets you specify how you would like commands completed (which happens when you enter the tab key). I suggest you place the following commands in your startup file:

   complete cd 'n/*/d/' # first arg of cd must be a directory
   complete 'dvips' 'n/*/f:*.dvi/' # only dvips the right files
   complete 'xdvi' 'n/*/f:*.dvi/' # only xdvi the right files
   complete 'xpdf' 'n/*/f:*.pdf/' # only xpdf the right files
   bindkey -v # use vi-style commands to edit command lines

3. You can establish useful command aliases. I use the following frequently:

   alias view 'vi +/\!\!:1/ `grep -l \!\!*`'	Edit files containing a given string.
   alias ls 'ls -F'	List files with suffixes showing type.
   alias lst 'ls -Llt \!\!* | less'	List files newest first.
   alias rm 'rm -i'	Remove files carefully.
   alias pd 'pushd'	Switch directory, saving old directory on a stack.
   alias nd 'mkdir \!\!$ ; cd \!\!$'	Create a new directory and switch to it.
   alias showtar 'gunzip -c < \!\!* | tar tf - '	Display the contents of a tar.gz file.
   alias showbtar 'bunzip2 < \!\!* | tar tf - '	Display the contents of a tar.bz2 file.
   alias maketar 'tar cf - \!\!* | gzip > \!\!$.tar.gz'	Create a tar.gz file.
   alias makebtar 'tar cf - \!\!* | bzip2 > \!\!$.tar.bz2'	Create a tar.bz2 file.
   alias untar 'gunzip -c < \!\!* | tar xf - '	Unpack a tar.gz file.
   alias unbtar 'bunzip2 < \!\!* | tar xf - '	Unpack a tar.bz2 file.

Music and video

• The mplayer program is capable of playing many kinds of audio and video streams, and it even has a plugin for browsers. It understands URLs as well as local file names.
  1. Local radio station WEKU: mplayer -cache 512 http://pubint.ic.llnwd.net/stream/pubint_weku
• The grip program can "rip" (convert from audio to MP3 or OGG or WAV) your CDs. It has access on the web to databases that store such items as artist name and track labels, so you usually don't have to enter this information, which is typically stored along with the sound itself in MP3 and OGG files.
• The xmms program is very good for playing your stored audio files.

Calculations

• For quick-and-dirty calculations, I recommend bc. It is capable of arithmetic with arbitrary precision (real numbers) and unlimited precision (integers).
  1. I usually give bc the command scale = 5 to compute and display real results to the nearest 0.00001.
  2. At its simplest, bc accepts expressions (one per input line) and prints their values. If you type in, for instance, 2^30/10^9, it prints out 1.07374 (to five decimal places).
  3. You can assign variables (lower case only) to values, as in a = 2^30-(5*2^20) and use them in later expressions, as in a - 2^14. (The result is 1068482560.)
  4. You can introduce functions, build loops, and do other more complex operations, but Perl programs are usually better for such purposes.
  5. If you need math routines, invoke bc -l, and then you can use functions like e() (exponentiation), l() (logarithm), and s() (sine).
• For spreadsheets, I recommend you either use gnumeric or soffice (OpenOffice).

Working from home or on the road

• If your home computer is running Microsoft Windows of any variety, you most likely want to connect to the department computers using putty, which is like ssh but builds its own window. I usually set it to use UTF-8 encoding and bold, larger font. A putty connection to a Unix machine gives you a command-line prompt, from which you can run programs like mutt, vim, and make.
• If your home computer is a Mac or running Unix, you can connect to department computers using ssh. Many cable and DSL providers will close your connection unless they think it is active. I suggest you place ServerAliveInterval 5 in your ~/.ssh/config file so that ssh sends an "are you there" message every 5 seconds.
• It is usually better to connect to coreopsis or penstemon, which have very high-quality connectivity, than to an office computer on the 7th floor of Anderson. You will find it far more responsive. However, mail sent from those two machines to uky.edu sometimes gets rejected.
• If your home computer is a Mac or a Unix machine running X-Windows, you can use ssh to connect to a department computer in a fashion that "tunnels" an X-Windows connection.
  1. You set up tunneling by invoking ssh -Y computerName. You can get this effect automatically by placing a line saying ForwardX11Trusted yes in your ~/.ssh/config file.
  2. Now you can start X-Windows clients such as xterm or gedit, and they will display directly to your home X-Windows server.
  3. If you use -X instead of -Y, you can still tunnel X-Windows connections, but these connections are "untrusted", meaning that they can't interact with other applications, such as with cut and copy.
  4. However, unless you have a high-bandwidth connection, you will find that running an X-Windows client remotely (on the department machine, displaying on your home machine) gives you very poor response. I have found that xdvi is usable in this mode on a DSL connection, but not many other programs. Cable connections do better than DSL.
• If you have a high-bandwidth connection and your home computer runs Linux, you can mount your department files into the file system of your home machine using sshfs. You can then navigate through your department files and you can read, write, and copy them as if they were on your home machine. There is a network delay, of course, as information is transmitted between home and department. To set up the connection, build a directory on your home machine, such as ~/uk/. Your department files will appear there when you run this command:

  sshfs -C -o idmap=user -o follow_symlinks
  	cor.cs.uky.edu: ~/uk

  If ssh usually prompts you for your department password, then sshfs will, too. To unmount the department files from your home machine, run this command:

  fusermount -u ~/uk

  You might want to introduce shell aliases for these two commands; I use sshuk and usshuk.
• You can establish a virtual screen on your department computer and connect to it using a virtual network computing (vnc) client. This method works acceptably on DSL lines, but is most likely not usable on dial-up lines.
  1. To start the server, on a department machine run

     vncserver :4 -localhost -geometry 1600x1400 -depth 24

     You might want to adjust the geometry (the size of the virtual screen) and the depth (the number of bits used to represent color). The :4 parameter says that the new virtual screen is the fourth X-Windows screen; it is very likely that this number works fine for you if the department machine is your own; you might need to find another number if you use a shared machine. You may set up the server at any time and just leave it running. It will ask you to establish a password. It will only accept connections from the department machine on which you start it.
  2. To run a client on your home machine, first start an ssh session from your home computer to the department machine where vncserver is running, tunnelling the vnc port:

     ssh -L5904:127.0.0.1:5904 machine.cs.uky.edu

     Then in a shell on your home machine, run

     vncviewer :4

     vncviewer will ask you for the screen password you set when you started vncserver.
  3. Better, if you have a recent version of vncviewer, you can run

     vncviewer -via machine.cs.uky.edu :4

     directly from your home machine; it will set up the ssh connection for you.
  4. The F8 key asks vncviewer to give you a menu, with useful options to close the viewer or to switch to full-screen mode.
• The University of Kentucky provides a Virtual Private Network (VPN) that you can access from your home or travel computer. The VPN allows you to act as if you are at the University. You can then access journals to which the University has a subscription. The VPN also circumvents the restrictions on some networks that prohibit ssh connections. Unfortunately, you can't connect to the VPN from on-campus wireless.
  1. You need the VPN client. On Ubuntu, use apt-get to install the network-manager-vpnc package.
  2. Configure VPN. Either run nm-connection-editor (as root) or click on the nm-applet icon in your control panel. You need to select "Cicso Compatible VPN (vpnc)" as the type of connection. The server address is vpn.uky.edu (or 128.163.55.43). The account name is your LinkBlue name. The group name is ukyedu. The group secret is also ukyedu.
  3. After you have established a connection (wireless or wired) to the internet, you can click on the nm-applet icon and choose the VPN network you have configured. You will get a prompt for your password (use your LinkBlue password) and maybe for the secret (it's still ukyedu).
  4. If you are using an Android device, install OpenConnect. Configure a new connection with server name vpn.uky.edu (or 128.163.55.43). Don't set up anything for CA certificate or authentication. "Software token" should be disabled. When you connect, give your LinkBlue user name and password.
  5. On an iOS device you can use Settings → General → VPN and establish the connection, using type IPSec, server 128.163.55.43, account your LinkBlue user name, group name ukyedu, secret ukyedu.
  6. When you start or stop the VPN, all current connections (such as for ssh) close. With the VPN running, traffic you send to computers at UKY will look like it comes from a private UKY IP address, like 172.31.18.206. Traffic to other computers will look like it comes from a UKY address, such as 128.163.237.38.

Printing

• Our printer software, called CUPS, allows you to print text, PostScript, and PDF files without first converting them. Just use lpr -Pprinter file ... If you always print to the same printer, you can set your PRINTER environment variable and then leave off the -Pprinter flag.
• If you want to print on a printer than has double-sided capacity, use the -d form of the printer name, as in csmain-d.
• If you wish to save paper, you can print multiple pages on a single side of paper by using the mpage program. For example, mpage -4 -P file > newfile.ps takes file and creates from it a newfile.ps PostScript file that represents the original file with pages reduced so that 4 fit on a single page. The -P parameter specifies that you don't want the result printed immediately but rather directed to standard out.

Virtual environments

• Sometimes you have Microsoft Win32 software that you simply must run, but you don't want to leave the world of Unix.
• You can set your machine to dual-boot, so you can decide as you boot whether you want Unix or a Microsoft operating system.
  1. If you want to let the two communicate with each other, you might want to configure your disk with a small FAT32 partition, which all Microsoft operating systems and Linux can read and write. You can put shared files in that partition.
  2. You can also use a flash drive as a way to communicate between the various operating systems.
• You can run many Win32 programs under wine. Just type

  wine myProgram

  1. The first time you run wine it will build the directory ~/.wine/, where it builds a virtual disk (called drive_c). If your program is an installer, it will install software in that virtual disk.
  2. You can later start the software by changing to the appropriate directory in the virtual disk and running wine on the software's .exe file.
  3. wine is not perfect; it implements most but not all the Win32 API, so your mileage may vary.
• You can use VirtualBox to create one or more virtual machines, each with its own disk (backed by a Unix file) and memory.
  1. It takes some trial and error (or expert assistance) to configure VirtualBox.
  2. You need an install CD for the guest operating system that you wish to load into each virtual machine. A disk image of the install CD (called an ISO image) also works.
  3. Once you have configured VirtualBox and installed a guest operating system in a virtual machine, you can start that machine, pause it, stop it, and save snapshots ("checkpoints") of its state.
  4. Each virtual disk and each snapshot uses lots of disk space. You should most likely configure VirtualBox to use local disk (if you have enough space there) instead of a directory mounted via NFS to hold these items.

Password maintenance

• It's not a good idea to use a single password for all your password-protected data, such as financial and consumer web sites. But if you have lots of different passwords, it is easy to forget them.
• One solution is to have a password-building pattern, where a few of the characters in the pattern pertain to the site in question. For example, your pattern might be 9!xrrtgy@, where x and y are site specific, the first two letters of the name of the site. For EBay, those letters would be E and B, giving the password 9!ErrtgB@. But if any one of your passwords gets revealed, they are all at risk if anyone can figure out the pattern. The method doesn't work for sites that force you to create new passwords every so often.
• Another solution is to have a password-protected keyring that contains all your other passwords.
  1. You can implement a manual keyring by building an encrypted text file and placing all your passwords there along with other information such as the web site and your user id at that site. The Vim text editor understands encrypting files; give it the :X command when you first edit the file to assign the file an encryption key; whenever you edit the file again, Vim will prompt you for the key. Be aware, though, that the encryption method Vim uses is not particularly secure. Your keyring file should also have file protections and perhaps an obscure name.
  2. There are programs that manage keyrings for you. I have been using gorilla, which has some nice features:
     1. It can generate good passwords for you. Since you don't have to remember them, the passwords can be truly obscure.
     2. You can organize your passwords according to category. I use categories for computers, financial sites, and commercial sites.
     3. You can click on an entry and thereby place the associated password into your cut buffer without ever actually seeing it on the screen, and when you paste it into a web form, it usually is hidden there, too, so people watching you never see the password you have entered into the form.