Back in February I wrote about using xli to add a desktop background of your choice to a minimalist window manager. I chose to write about xli for two reasons. First, several window manager developers choose to use xli by default. For example, if you look at a .jwmrc file, the configuration file used by JWM, a lightweight window manager I am rather fond of, you will see that xli is used in the <Startup Command> section. The second and perhaps more important reason I chose to write about xli is because it’s what I knew and used for years. One thing about Linux and UNIX: there are always different ways to do things. It turns out that many distros include something a bit newer and perhaps better than xli.

Esetroot can also be used to change the contents of the root window in X. The root window is your desktop background. The advantage of Esetroot over xli is that it supports transparency in applications. This is a low resource piece of eye candy that I particularly like in terminal emulators like mrxvt, aterm, and xfce Terminal. I like seeing my background, albeit shaded, perhaps in a color of my choice, in the background of my terminal window. pypanel, a small panel or toolbar application for minimalist window managers written in Python, also supports transparency nicely.

Mac rather than Linux, but never mind.

One of my users recently had his PowerBook fail, and needed to reinstall from an external source (as the DVD drive was broken). Apparently all the information he found online insisted that this could only be done via FireWire, either from an external DVD drive or by booting in Target mode (attached to another Mac). However, after some days of prodding at it, he eventually tried a USB external DVD drive, and that worked absolutely fine, albeit slowly.

While I’m on the subject, consider this a reminder to back your data up. I say this because my own laptop died last week (thankfully I did have a recent backup!). The Mac Target mode did come in handy here. My hard drive was showing errors when I booted verbosely (hold down Apple-V as you press the power button & hold it until the boot screen appears - may take some time). However, when I attached it to another Mac via FireWire, and booted it in Target mode (hold down Apple-T during boot), it (eventually…) showed up as a volume on the other Mac. This meant I was able to rescue the one really important file that hadn’t been backed up in its most recent version, as I’d been working on it that day. After which, sadly, it expired entirely. I do now have a lovely shiny new black MacBook (prioritising size over the increased power of the PowerBook - it’s a shame they no longer make 12″ PowerBooks) as a replacement.

On July 22nd a new set of kernel packages was released for Vector Linux, my chosen primary and current favorite distribution. This was the second build of the 2.6.21 kernel with Con Kolivas’ CK2 patchset, replacing a test build released on July 8. In the past the only reason I’ve recommended upgrading a kernel is to close security vulnerabilities or to add support for new hardware. Recently, though, there is another very good reason: noticeably improved performance, particularly if you are currently using kernel 2.6.19 or earlier.

At OSCON a few minutes ago, Nat Torkington asked Bill Hilf what will change in the future specifically with regard to Microsoft dealing with F/OSS projects regarding patents.

(The context is that Hilf had just said that Microsoft has made mistakes in certain areas.

Hilf’s response was… priceless. “I get a lot of e-mail.” “People like to subscribe me to crazy newsletters and spam.”

Uh, thanks.

Hi! I’m Kevin Farnham, a career software developer, and sometimes editor and writer. Most of my software development work has been on Unix/Linux platforms, but I also have significant experience on Windows (especially on the server side). My core experience is in the areas of complex scientific programing (mathematical modeling and simulation) and high volume data processing and data access. I’ve worked with multiprocessor systems, and developed multithreaded applications, for a very long time.

As you can see from the articles I’ve published on the O’Reilly Network (click the Articles tab on my O’Reilly Network profile page to see the articles list), I’ve worked with quite a few Linux distributions. My current favorite is Gentoo — because it lets me have exactly what I want to have in my system, and nothing more. I use XFCE as my desktop environment. Again, it gives me what I need, without extraneous clutter.

I’ve just started working on a new project involving Intel’s Threading Building Blocks, a C++ template library that simplifies development of multithreaded software for use on Linux, Windows, and Mac operating systems. With multicore processors quickly becoming the norm, software development is going to have to change — otherwise, applications will utilize only a small portion of the available processing power on modern PCs. I’ll be posting quite a lot about that specific topic area in my blog on the Intel Software Network Blogs site.

As for other interests: I often find myself “overly” fascinated by subjects that would appear to many to be obscure, such as how bootloaders work, the poetry of the Middle English author known as “the Pearl Poet”, the significance of quantum mechanics, algorithms, and Samuel Taylor Coleridge’s philosophy of Nature. Right now, in addition to my studies of Threading Building Blocks, I’m reading “Causality and Chance in Modern Physics” by David Bohm, and I’m rereading John Milton’s “Paradise Lost”. All quite good brain exercise, I think!

What’s up next for me is: OSCON! I’ll be attending next week for the first time. It will also be my first visit to Portland, Oregon. When I’m not attending tutorials or sessions, you’ll probably be able to find me hanging out at the Intel and O’Reilly booths, or somewhere nearby. Let me know if you’d like to connect.

And I’ll be sure to post here, from there.

Thanks to everyone who commented on The Dubious Benefits of Porting F/OSS to Windows. There are a lot of good points in the comments.

For example, Simon Hibbs responded that more users of a piece of software increased the value of that software. This is important for programs that follow open standards, such as Mozilla Firefox, or OpenOffice.org and Abiword. In one sense, I see the value of open standards and unfettered access to data as more important than the four software freedoms. If I had to choose one over the other, I’d choose open standards… but I think that’s a false dilemma.

Yet I still wonder. How many people have switched to free platforms after realizing that they already used or could switch to completely free software that did what they needed?

I’ve seen plenty of people switch from a free OS to a proprietary system which includes some free parts (let’s call it “Mac Something Something”… or how about “Something OS X” to protect the innocent), justifying that choice by saying “It can run all of the applications I love from Linux or *BSD, and look it’s shiny!”

I’m not sure that that case really helps the goal of spreading free software.

Again, I realize that not everyone shares that goal with me, and it’s fine. I’m also not saying that porting free software to proprietary platforms is bad, or wrong, or makes you an evil person. I’m sympathetic to the idea that plenty of people develop on non-free platforms and deploy to free platforms. In my last full-time development job, I would have gone crazy if not for Cygwin. (I ended up using its X server to connect to our GNU/Linux test machine just so I could use decent development tools.)

Yet still I wonder… is there concrete evidence that people do switch to free platforms after using primarily free software? Is there concrete evidence that open standards gain significant acceptance due to the presence of free software or software that otherwise promotes those standards effectively? (I’d like more than one case; it’s possible that Mozilla is an outlier.)

(I haven’t brought up the case of writing free software that supports only a proprietary standard, but that’s a different post altogether.)

Thanks to everyone who commented on my last post, about Jabber servers. Quite a few of them seem to play nicely with LDAP; the trouble with the setup here is that we have LDAP + Kerberos, and it’s the Kerberos part that seems problematic. There is of course the “roll your own” option but I am currently preferring to steer clear of that in favour of accepting that users will have to create themselves a separate Jabber account rather than having everything tie together neatly. It’s not terribly difficult to do, after all!

A couple of things to bring to your attention: firstly, the usefulness of cron-apt (for apt-using systems) is probably well known already, but since I only encountered it about a year ago, I commend it to anyone else who hasn’t done so yet. It runs a daily cron-job to check for package updates for you. I recommend setting it to download only, not to automatic install, and also to only email if there’s an update waiting. To actually run the upgrades once you’ve checked the email to ensure it’s not done anything foolish, you can then use either ClusterSSH to run apt-get upgrade, or some form of roll-your-own solution, such as this:

#!/bin/sh
# Script to automate apt-get run across machines
# Usage: deb_aptget.sh "machine1 machine2 machine3"

case $@ in
"") echo "Usage: `basename $0` \"machine1 machine2 machine3 ...\""
        exit 1
        ;;
esac

# Authenticate root passphrase
ssh-add .ssh/root_key

# Run command on each of listed machines
for arg
        do
                echo $arg
                ssh root@$arg -i .ssh/root_key apt-get -y dist-upgrade
        done

# Deauthenticate root
ssh-add -d .ssh/root_key

Customary warnings apply, in particular the fact that I am not responsible if you use this & do something dramatically awful to your system(s).

Secondly, you may be interested in this article about OpenGuides, a Perl-based wiki that allows people to build open-source guides to cities. I confess to being biased, since I’ve been a contributor (on the content, rather than the code, side) to a couple of OpenGuides wikis (& know the author of the article!), but it is a very interesting project and well worth a look.

A couple of weeks ago, the O’Reilly editors asked Is Microsoft Relevant in a Post-Rails World? Contrary to some reports, there are still desktop applications in the world besides a web browser–and there are plenty of desktop applications under serious development.

Many of them are F/OSS. Of those, plenty have ties to existing projects to produce fully-free desktop software. They run on top of free Unixes, take advantage of free APIs and libraries, and interoperate well with other free software.

There are very few technical reasons much of this software cannot run on non-free platforms, however. One of the first pieces of free software I used reliably was Emacs (though I quickly switched to Vim). This was on Windows NT 4. Perhaps the most successful free software desktop application is Mozilla Firefox.

A request for opinion/experience today: does anyone have experience of running (local) Jabber servers under Linux? (Debian, ideally).

I’ve been experimenting with setting up a Jabber server to run within our LAN. I tried Openfire and found it quite hard work. Ejabberd worked better, but after running for a few hours on the same machine as our webserver, there were a lot of zombie blosxom.cgi processes and the webserver was no longer responding. I haven’t definitely confirmed this behaviour as due to ejabberd, but am not terribly keen to experiment further on a live webserver. I will be trying it on another machine to see how that behaves; but in the meantime, a) has anyone else seen problems on a server running ejabberd? and b) can anyone recommend a Jabber server that is straightforward to set up, well-behaved once running, and supports chatrooms?

I would also very much like for it to hook into our LDAP/Kerberos setup. I gather that single-sign-on is a nonstarter (since gaim and other clients don’t do GSSAPI), but being able to use the same usernames and passwords would be very useful. In theory both Openfire and ejabberd can be made to do this; in practice IME it doesn’t work. Any thoughts on that would also be welcome!

I’m making the leap from point-and-shoot digital cameras to digital SLRs. This represents a considerable increase in cost and hassle. The latest generation of compact point-and-shoots are pretty darned good. Will it be worth it? Or will I be smitten with terminal buyer’s remorse? Does Linux offer sufficient high-end photo management and editing software? Visit Adventures in Digital Photography With Linux, part 2 to find out. The first two installments cover cameras and lenses, and digital arcana thereof. Over the next few weeks I’ll write about editing and managing photos on Linux.

And, of course, you fine readers are invited to share your own Linux photography adventures.