Command-line programs

Bash – the Bourne-Again Shell – appears on a Linux system whenever you open a terminal window. Normally Bash is used for one-off commands, but it is possible to string these together into long and powerful programs including parameters and internal variables.

Bash programs are created as text files with the first line

#!/bin/sh

This is usually followed by a line specifying the script name. The file can be saved with the extension .sh, though this is not essential.

This simple example combines all the PDF files in a directory, extracts the text from the resulting file, and counts the words in that file:

#!/bin/sh
# Script Name: combinepdfs.sh
pdftk ./*.pdf cat output ./complete.pdf
pdftotext complete.pdf
wc -w complete.txt >>wordcount.txt

To run a Bash script, right-click it in the Nautilus File Manager and select Properties. Click on the Permissions tab and turn on the check box which says Allow executing file as program. You should then be able to run the script by double-clicking on it. Alternatively you can open a terminal window, navigate with cd to the directory containing the file, and type the name of the file to run it.
Here are four that I find particularly useful.

Eject a CD, pause ten seconds and close the CD drawer

su
umount -l /media/cdrom0eject /media/cdrom0
sleep 10
eject -t

Split all mp3s in a folder into ten-minute segments

Handy for breaking up long music podcasts (like Ultima Thule! – www.ultimathule.info) into convenient sections.

mp3splt -f -t 10.0 -a -d 10split UT*.mp3

Combine all PDFs in a folder

The command pdftk won’t work if the PDF filenames contain spaces, so the first line replaces all spaces in the names with underline characters, and the third line changes them back again:

rename -v 's/\ /\_/g' *.pdf
pdftk *.pdf cat output complete.pdf
rename -v 's/\_/\ /g' *.pdf

Backup

A simple backup script that synchronises the contents of my /home directory with an external hard disk called Transcend:

#/bin/bash
sudo mount -a
rsync -v -r -l --delete -z /home/jon /media/Transcend
echo "Finished!"

The Distrowatch Top Ten

Every day the independent website Distrowatch updates its list of Linux distro page hit rankings. This is, of course, not the same as a popularity contest: the fact that many people are looking at a page about Distro X doesn't mean that all those people are installing X or using X, or even thinking about it, but in the absence of any reliable figures (one of the drawbacks of free distribution), the Distrowatch figures are the best indication we have of the relative standing of each distro. The time period for the rankings is adjustable, so it is possible to do a bit of crude trend analysis and estimate where the figures might be going as well as seeing where they have been. As an ex-statistician, it's interesting for me to see just what I can deduce from all this.

Here are the Distrowatch figures for the last three years, with a couple of columns of analysis. The figures show the average number of hits per day over the period of a year for each of the ten distros (the actual table goes down to 100 distros, but I'll spare you those). The table includes links to the relevant distro pages on the site. If the figures are a bit small, hold down the Control Key and spin your mouse wheel -- that should zoom you in.

What does this all tell us? Firstly, there is a remarkable consistency across the three years. Eight of the distros that were in the top ten in 2007 were still there in 2009. Since the detractors of Linux like to point out how fickle and evanescent the Linux world is, it's nice to have some confirmation that things really don't change that much.

Secondly, this confirms the ongoing dominance of Ubuntu. Despite going through six new versions in the three-year period, Ubuntu has maintained its dominance. Its main competitor in 2007, PCLinuxOS, has dropped to eighth place in 2009, largely because of its failure to adhere to a rigid release schedule (translation: the new version was late). Fedora has picked up significantly with its new release in 2009, but still barely attains two-thirds of the hits Ubuntu does. Add in Mint, which is only a minor variant of Ubuntu, and the system gets 31% of the daily hits in the Top Ten, more than twice those of any other system.

It's interesting that the number of hits on Ubuntu has actually dropped over the last two years. It certainly isn't because of declining interest, so I'm going to put that down to a widening range of other websites (like this one) where Ubuntu fans and the curious can go to get their fix of information. If anyone has a better idea, let me know.

Particularly interesting is the drop by OpenSUSE from second to fourth place between 2008 and 2009. SUSE is a KDE-based system which has acquired a wide and enviable reputation for being reliable and bulletproof. Reviews of SUSE may describe it as 'boring' but never 'unreliable', Despite this it has failed to capture the imagination, and its hit rate is declining rapidly. Is this a reaction to the new version of KDE? Is GNOME now 'winning' the interface contest, with four GNOME-based distributions in the top five? Or will there be a swing back to KDE systems as the new KDE version 4 is tested and debugged more extensively?

What does this all mean for beginners? Simply that, unlike a Windows system, using Linux doesn't lock you into a particular person's (or company's) philosophy. All the distros above (and the other 298 currently listed on Distrowatch) represent the particular take that a certain individual or small group has on Linux and what it should do. If you happen to find the distro which matches your needs it can be like finding your soulmate: everything works and it all seems right. If not, don't despair: there are lots more to try, and they're (nearly) all free. Using Windows is like putting your hand into a vice: using Linux is like putting it into a lucky dip.