9. The Remaining 1%

More than 1%, actually...

9.1 Using tar & gzip

Under UNIX there are some widely used applications to archive and compress files. tar is used to make archives---it's like PKZIP but it doesn't compress, it only archives. To make a new archive:

$ tar -cvf <archive_name.tar> <file> [file...]

To extract files from an archive:

$ tar -xpvf <archive_name.tar> [file...]

To list the contents of an archive:

$ tar -tf <archive_name.tar> | less

You can compress files using compress, which is obsolete and shouldn't be used any more, or gzip:

$ compress <file>
$ gzip <file>

that creates a compressed file with extension .Z (compress) or .gz (gzip). These programs can compress only one file at a time. To decompress, use:

$ compress -d <file.Z>
$ gzip -d <file.gz>

RMP.

There are also the unarj, zip and unzip (PK??ZIP compatible) utilities. Files with extension .tar.gz or .tgz (archived with tar, then compressed with gzip) are as common in the UNIX world as .ZIP files are under DOS. Here's how to list the contents of a .tar.gz archive:

$ tar -ztf <file.tar.gz> | less

9.2 Installing Applications

First of all: installing packages is root's work. Most Linux applications are distributed as a .tar.gz archive, which typically will contain a directory called pkgname/ containing files and/or subdirectories. A good rule is to install these packages from /usr/local with the command

# tar -zxf <archive.tar.gz>

reading then the README or INSTALL file. In many cases, the package is distributed in source, which you'll have to compile to create the binaries; often, typing make then make installwill suffice. Obviously, you'll need the gcc or g++ compiler.

Other archives may have to be unpacked from /; this is the case with Slackware's .tgz archives. Other archives contain the files but not a subdirectory. Always list the contents of the archive before installing it.

The Debian and Red Hat distributions have their own archive format; respectively, .deb and .rpm. The latter is gaining wide acceptance; to install an .rpm package, type

# rpm -i package.rpm

9.3 Tips You Can't Do Without

Command completion: pressing <TAB> when issuing a command will complete the command line for you. Example: you have to type gcc this_is_a_long_name.c; typing gcc thi<TAB> will suffice. (If you have other files that start with the same characters, supply enough characters to resolve any ambiguity.)

Backscrolling: pressing <SHIFT + PAG UP> (the grey key) allows you to backscroll a few pages, depending on how much video memory you have.

Resetting the screen: if you happen to more or cat a binary file, your screen may end up full of garbage. To fix it, blind type reset or this sequence of characters: echo CTRL-V ESC c RETURN.

Pasting text: in console, see below; in X, click and drag to select the text in an xterm window, then click the middle button (or the two buttons together if you have a two-button mouse) to paste. There is also xclipboard (alas, only for text); don't get confused by its very slow response.

Using the mouse: if you installed gpm, a mouse driver for the console, you can click and drag to select text, then right click to paste the selected text. It works across different VCs.

Messages from the kernel: have a look at /var/adm/messages or /var/log/messages as root to see what the kernel has to tell you, including bootup messages. The command dmesg is also handy.

9.4 Where to Find Applications

If you're wondering whether there are applications to replace your old DOS/Win ones, I suggest that you browse the main Linux software repositories: ftp://sunsite.unc.edu/pub/Linux , ftp://tsx-11.mit.edu/pub/linux , and ftp://ftp.funet.fi/pub/Linux . Another excellent place is the ``Linux Applications and Utilities Page'' at http://www.xnet.com/~blatura/linapps.shtml .

9.5 A Few Things You Couldn't Do

Linux can do an awful lot of things that were cumbersome, difficult or impossible do to with DOS/Win. Here's a short list that may whet your appetite:

• at allows you to run programs at a specified time;
• awk is a simple yet powerful language to manipulate data files (and not only). For example, being data.dat your multi field data file,

  $ awk '$2 ~ "abc" {print $1, "\t", $4}' data.dat
  

  prints out fields 1 and 4 of every line in data.dat whose second field contains ``abc''.
• cron is useful to perform tasks periodically, at specified date and time. Type man 5 crontab.
• file <filename> tells you what filename is (ASCII text, executable, archive, etc.);
• find (see also Section Directories: Translating Commands) is one of the most powerful and useful commands. It's used to find files that match several characteristics and perform actions on them. General use of find is:

  $ find <directory> <expression>
  

  where <expression> includes search criteria and actions. Examples:

  $ find . -type l -exec ls -l {} \;
  

  finds all the files that are symbolic links and shows what they point to.

  $ find / -name "*.old" -ok rm {} \;
  

  finds all the files matching the pattern and deletes them, asking for your permission first.

  $ find . -perm +111
  

  finds all the files whose permissions match 111 (executable).

  $ find . -user root
  

  finds all the files that belong to root. Lots of possibilities here---RMP.
• grep finds text patterns in files. For example,

  $ grep -l "geology" *.tex
  

  lists the files *.tex that contain the word ``geology''. The variant zgrep works on gzipped files. RMP;
• regular expressions are a complex but darn powerful way of performing search operations on text. For example, ^a[^a-m]X{4,}txt$ matches a line that starts with `a', followed by any character except those in the interval a-m, followed by 4 or more `X', and ends in `txt'. You use regular expressions with advanced editors, less, and many other applications. man grep for an introduction.
• script <script_file> dumps the screen contents on script_file until you issue the command exit. Useful for debugging;
• sudo allows users to perform some of root's tasks (e.g. formatting and mounting disks; RMP);
• uname -a gives you info about your system;
• The following commands often come in handy: bc, cal, chsh, cmp, cut, fmt, head, hexdump, nl, passwd, printf, sort, split, strings, tac, tail, tee, touch, uniq, w, wall, wc, whereis, write, xargs, znew. RMP.

9.6 Common Extensions and Related Programs

You may come across scores of file extensions. Excluding the more exotic ones (i.e. fonts, etc.), here's a list of who's what:

• 1 ... 8: man pages. In the unlikely case you don't yet have it, get man.
• arj: archive made with arj.
• dvi: output file produced by TeX (see below). xdvi to visualise it; dvips to turn it into a PostScript .ps file.
• gz: archive made with gzip.
• info: info file (sort of alternative to man pages). Get info.
• lsm: Linux Software Map file. It's a plain ASCII file containing the description of a package.
• ps: PostScript file. To visualise or print it get gs and, optionally, ghostview or gv.
• rpm: Red Hat package. You can install it on any system using the package manager rpm.
• taz, tar.Z: archive made with tar and compressed with compress.
• tgz, tar.gz: archive made with tar and compressed with gzip.
• tex: text file to submit to TeX, a powerful typesetting system. Get the package tex, available in many distributions; but beware of NTeX, which has corrupted fonts and is included in some Slackware versions.
• texi: texinfo file, can produce both TeX and info files (cp. info). Get texinfo.
• xbm, xpm, xwd: graphic file. Get xpaint.
• Z: archive made with compress.

9.7 Converting Files

If you need to exchange text files between DOS/Win and Linux, beware of the ``end of line'' problem. Under DOS, each line of text ends with CR/LF, while under Linux with LF. If you try to edit a DOS text file under Linux, each line will likely end with a strange--looking `M' character; a Linux text file under DOS will appear as a kilometric single line with no paragraphs. There are a couple of tools, dos2unix and unix2dos, to convert the files.

If your files contain accented characters, make sure they are made under Windows (with Write or Notepad, say) and not under plain DOS; otherwise, all accented characters will be screwed up.

To convert Word or WordPerfect files to plain text, the matter is a bit trickier but possible. You'll need one of the tools that can be found on the CTAN sites; one is ftp://ftp.tex.ac.uk . Get the package word2x from the directory /pub/tex/tools/', or try one the packages available in directory /pub/tex/support/. I've only tried word2x, and works quite well.