Productive Monday

Fedora uses a worldwide system of mirror servers to distribute packages. When a Fedora system needs package info, packages, or updates, it will by default request a mirror list from a Fedora server. This list is generated based on the repository and architecture requested as well as the IP address of the requesting system.

The yum-fastestmirror package provides a Yum plugin which measures the speed of available mirrors and sorts the list so that priority is given to the fastest mirrors; this can result in faster yum operations and reduced network congestion -- and with no additional manual effort beyond installing the plugin.

Installing from the command line: yum install yum-fastestmirror
Installing using the graphical installer: Not available by browsing; search for yum-fastestmirror
Menu location after installation: Not on the menu; alters the operation of yum and PackageKit
Command: No direct command; access functionality via /usr/bin/yum and /usr/bin/pkcon
Upstream website: http://yum.baseurl.org/

The shell has gradually improved through the years, from the Multics shell to the Bourne shell (sh), Korn shell (ksh), Bourne-again shell (bash), and other derivatives (ash, zsh). There are shells with C-like syntax (csh, tcsh) and others. But most shells retain a text-based paradigm and a scrolling-text display.

Hotwire is an object-oriented "hypershell" with an integrated GUI that offers a different way to work. Shell builtins such as "ls", "proc", and "filter" are used to build object pipelines, and the GUI displays pipeline results in an intelligent manner, allowing you to sort by clicking on column headings, change directories by double-clicking on paths, and so forth. Hotwire enables you to interactively extend pipelines by taking the current output and (without re-execution) feed it into additional commands. The commands are object-aware and an object inspector built into the GUI provides easy access to property and method details.

In addition to builtin commands, hotwire allows you to run Python snippets and regular shell commands (such as loops). It also provides TAB-completion and history searching, evaluates Perl and Ruby expressions, searches command output, and can run commands in a traditional terminal emulator.

This project has come a long way in just over a year, and it will be interesting to see how it progresses as planned features such as remoting are added.

Installing from the command line: yum install hotwire
Installing using the graphical installer: Not available in the Browse view; use the List or Search view to install hotwire
Menu location after installation: Applications > System Tools > Hotwire Shell
Command: /usr/bin/hotwire
Upstream website: http://hotwire-shell.org/

Meld is a powerful visual diff and merge tool. It displays colour-coded two- and three-way diffs and enables you to merge or edit the compared files. It can work with version control systems (cvs, svn, hg) and can diff directories. The GTK-based interface is clean and features tabs and a toolbar.

Meld does a great job of simplifying edit and merge tasks; when comparing files, for example, you can simply click on the arrows in the middle column to merge text from one side to the other (see screenshot). Holding down Ctrl reveals insert-before/insert-after options, and holding down Shift enables one-click block deletion.

Installing from the command line: yum install meld
Installing using the graphical installer: Development > Development Tools > meld
Menu location after installation: Applications > Programming > Meld Diff Viewer
Command: /usr/bin/meld
Upstream website: http://meld.sourceforge.net/

Thanks to Alessandro Boggiano for suggesting Meld!

If you use more than one computer in your daily work -- especially if one is a laptop -- it's quite common to end up with a mismash of files and file versions spread across the machines. Unison is an easy-to-use, GUI-based, bidirectional file synchronization tool. It runs on Windows and most flavours of Unix/Linux, so it enables cross-platform synchronization, and SSH tunneling provides effective security for data in transit without requiring additional key or firewall configuration.

Installing from the command line: yum install unison
Installing using the graphical installer: Base System > System Tools > unison
Menu location after installation: Applications > Accessories > Unison File Synchronizer
Command: /usr/bin/unison
Upstream website: http://www.cis.upenn.edu/~bcpierce/unison/

Thanks to Rahul Sundaram for suggesting Unison! If you'd like to suggest a package, please use the form on the right side of the main page.

Planet is a feed aggregator which assembles multiple RSS or Atom newsfeeds into a single web page and feed, like Planet Fedora, Planet Gnome, or Planet KDE. I use it to generate the opensource@seneca Planet.

To keep resource usage to a minimum, Planet writes static web pages based on template files instead of running as a web script. When the Fedora Planet package is installed, sample template files are installed in /usr/share/doc/planet-2.0/examples/, and CSS and image files for the sample templates are in /usr/share/doc/planet-2.0/examples/output.

To set up a planet, copy the template, CSS, and image files to appropriate locations (or create your own template files using the samples as a guide). In this example, I'm going to use ~/planet/template for the template files and ~/public_html/planet for the output location (where the CSS, images, and generated HTML/XML will be placed):

$ mkdir -p planet ~/public_html/planet
$ cp /usr/share/doc/planet-2.0/examples/*tmpl /usr/share/doc/planet-2.0/examples/fancy/*tmpl ~/planet
$ cp -r /usr/share/doc/planet-2.0/examples/output/* ~/public_html/planet

The next step is to create a configuration file. Again, examples can be found under /usr/share/doc/planet-2.0/examples/; here is a simple config file based on the file locations above:

# Configuration file for planet

# Basic planet details
[Planet]
name = myPlanet
link = http://concord3.proximity.on.ca/~chris/planet/
owner_name = Chris Tyler
owner_email = chris.tyler@example.com

# Directory for cached feed data
cache_directory = /home/chris/planet/cache

# Directory for output HTML and XML files
output_dir = /home/chris/public_html/planet

# List of template files to be processed
template_files = planet/template/atom.xml.tmpl planet/template/foafroll.xml.tmpl planet/template/index.html.tmpl planet/template/opml.xml.tmpl planet/template/rss10.xml.tmpl planet/template/rss20.xml.tmpl

# Feeds to be processed
[http://blog.chris.tylers.info/index.php?feeds]
name = Chris Tyler
face = images/chris_tyler_hackergotchi.png

[http://planet.fedoraproject.org/atom.xml]
name = Planet Fedora

[http://rss.slashdot.org/Slashdot/slashdot]
name = Slashdot

[http://dailypackage.fedorabook.com/index.php?feeds]
name = Fedora Daily Package

Save this file in an appropriate location (e.g., ~/planet/config.ini) and then run planet with this file as the only argument:

$ planet ~/planet/config.ini

If the output looks good, use crontab to schedule planet to run periodically:

$ (crontab -l;echo "50 * * * * /usr/bin/planet $HOME/planet/config.ini")|crontab

Installing from the command line: yum install planet
Installing using the graphical installer: Not available in the Browse view; use the Search or List views to install planet
Menu location after installation: Not applicable -- run from the command line or as a cron job
Command: /usr/bin/planet
Upstream website: http://planetplanet.org

htmldoc is a powerful simple-to-use tool which converts HTML to Postscript, PDF, or indexed HTML output. It provides a graphical user interface (pictured) for manual document conversion -- useful for tasks such producing printed manuals from web pages -- but it can also be used as a filter.

When used as a filter, htmldoc provides a simple way to produce nice-looking output from a script. Here's a bash example:

echo "<html><body><h1>Heading</h1><p>Some HTML output...</p></body></html>" |
htmldoc --format ps3 --bodyfont Helvetica --header ... --footer ... --webpage - | lpr

htmldoc is also a handy tool for producing PDF output in CGI scripts.

Since tables and images are supported, it's easy to set up documents such as invoices, statements, and reports with columnar data, logos, and icons. With a bit of sed/awk/perl magic, you can transform the text output of normal commands into HTML for inclusion in the htmldoc output.

This software is developed by Easy Software Products, who also develop the CUPS print spooler system used in Fedora.

Installing from the command line: yum install htmldoc
Installing using the graphical installer: Applications > Office/Productivity > htmldoc
Menu location after installation: Office > HTMLDOC
Command: /usr/bin/htmldoc
Upstream website: http://www.htmldoc.org/

In the late 1960's, a computer algebra system named Macsyma was developed at MIT. This system was commercialized by Symbolics, and a number of other programs such as the proprietary Maple and Mathematica systems were inspired by it.

Fedora includes the maxima package, an open-source descendant of the original Macsyma code (forked in 1982 and placed under the GPL in 1998). It provides powerful symbolic and numeric manipulation and 2D and 3D plotting capability, and the maxima-gui package provides the simple but functional xmaxima graphical front-end.

(Note: sometimes, an sbcl (Steel Bank Common Lisp) process will continue to run after Maxima has been terminated, consuming a lot of CPU time).

Installing from the command line: yum install maxima-gui
Installing using the graphical installer: maxima-gui is not available through the Browse view (although maxima is in the Engineering and Scientific group); use the Search or List views to install the maxima-gui package
Menu location after installation: Applications > Programming > xmaxima
Command: /usr/bin/maxima-gui (graphical interface) or /usr/bin/maxima (character-mode interface)
Upstream website: http://maxima.sourceforge.net/

Update: maxima package maintainer Rex Dieter points out that wxMaxima provides a nice alternate graphical front-end to maxima. (Thanks, Rex!)

System administrators have always relied on top for per-process CPU and memory usage statistics, and vmstat (or sar) to analyze I/O -- but there has been no easy way to analyze disk I/O on a per-process basis.

iotop is the missing tool. It shows the disk read and write rate, swapins, and total disk I/O for each process. The process list is sorted by I/O and updated once per second.

A minimal number of interactive controls are available: r reverses the sort order and q quits.

Although no manpage is provided for iotop, executing it with the argument --help reveals a few available options, including -d to set the delay between updates.