Click for 1680×1050 version

Sure, there are other ways to get the data I’m looking for, but having text rendered directly on the desktop like that just looks damn cool to me. It also intimidates the crap out of non-geeks when they see my office, which I fully admit is a big plus.

I came across a package called Devil’s Pie that gave me a lot more control over windows. Although I was looking for something that would let me actively select a window to give it certain properties (sticky, no decorations, etc), Devil’s Pie instead passively detects new windows and applies settings to them. It’s a mentality shift from my initial thoughts, but I think over time this model will actually be really useful. Plus, it has “pie” in the name, which is just awesome.

With a combination of gnome-terminal profiles/flags and Devil’s Pie configuration, I was able to get my desktop to automatically populate my desktop with all sorts of data. The following description of how is kinda wordy, but it’s really not complicated at all.

Install and Configure Devil’s Pie

sudo yum -y install devilspie
mkdir ~/.devilspie
touch ~/.devilspie/console.ds

Devil’s Pie reads all configuration files (I believe they have to end in .ds, but don’t quote me on that) in the above directory on startup. Those files define the rules that will be applied to new windows that are created (there’s also a way to ask it to apply rules to all existing windows as well). For this example, I put my rules in a file named console.ds.

(if 
(contains (window_name) "console" )
(begin 
(undecorate)
(pin)
(skip_pager)
(skip_tasklist)
)
)

There are two main parts to the above configuration. The first is the matching expression to decide if a window should have the rules applied. In this case, I indicate to apply the actions if the window name contains the string “console” (there are also options for “is” and “matches”). I’ll cover where that rule comes into play in a bit.

The actions do the following:

• undecorate – This is primarily what I was going for, I want the window decorations removed. This includes the title bar and border, though you can still use the shortcut key (alt+F8 by default) to resize a window without decorations.
• pin – Make this window available on all workspaces. There are also attributes to control putting a window on a particular workspace, which could be useful in my case since I use 12 workspaces and am, well, pretty anal retentive about having certain applications in certain workspaces.
• skip_pager – Don’t let the window show up in the pager. It removes the clutter in my pager since I don’t consider this a “real” application, just a status-type window.
• skip_tasklist – A little more useful than skip_pager, this keeps the window off the taskbar, for the same rationale as above.

Needless to say, I’m only scratching the surface of all of the actions supported. I haven’t found a good way to find this list yet; the man page says the -l flag will list symbols but when I try it, it just doesn’t do anything.

Run Devil’s Pie

devilspie

Not much to say about that. I’ll talk more about enabling a debug mode later, but I mention it now to say it’s not done through flags. You’ll need to restart it every time you change your confguration files, but once you’re set add it to System -> Preferences -> Startup Applications and you’re good to go on reboots.

Create Gnome Terminal Profile

To summarize above, Devil’s Pie will take care of removing window decorations and keeping it on the visible workspace. We still need to configure Gnome terminal to add some slickness to how it looks. Since I don’t want this slickness on all of my terminals, I created a new profile named “console” with the following changes:

• Background – Transparent background, with the slider set all the way to none.
• Scrolling – Scrollbar is: Disabled

That gives us a completely transparent shell, reflective of its use to simply throw data at me.

Start the Terminal

Now we just need to make sure we start the terminal such that it will use the right profile and will trigger the window detection from Devil’s Pie. Here’s where the matching expression from above comes into play. I don’t want all of my terminals to have no decorations, just specific ones I want to treat in this fashion. When starting the terminal, tell it to use a specific window title that will match our Devil’s Pie rule from above.

gnome-terminal -t console --profile=console --geometry=140x20+1680-0 -e htop

A quick explanation of above:

• -t – Sets the title of the terminal window. This value should match up to the window_name value from the Devil’s Pie configuration.
• –profile – Tells the terminal to use the slick console profile in all of its transparent goodness.
• –geometry – Initially size and place the terminal on my screen.
• -e – Run a specific command when it starts, in this case htop.

My screenshot above simply uses a similar call to gnome-terminal, passing a tail command to -e and different geometry to get the log tail on the top of the screen.

Two other notes on Devil’s Pie:

There is a UI for editing the configuration files: gdevilspie. It’s handy because you can easily see the different types of matching expressions and actions, however the files it generated didn’t run for me. For some reason, even with one matching expression it still added a blank condition and an “and” statement and closed the conditions block too early. I haven’t looked to closely yet to see what is wrong.

Enabling debugging is as simple as adding another configuration file. I added one named “debug.ds” to my ~/.devilspie directory that contained simply:

(debug)

Once that is in place, running the devilspie executable from the command line output some really useful window information whenever a new window was created.

After experimenting with a few different approaches, we found ourselves regularly using gobby. I took the post title directly from their site, but it doesn’t do justice to just how fluidly it all comes together.

Gobby lets everyone logged into a gobby server type at the same time on the same document. Ugh, even that doesn’t showcase how cool it works in practice. The changes are sent in real time to every user, which both provides a very seamless collaboration environment as well as lets my teammates know just how bad I really am at typing (I’m even worse when I realize other people are watching me mistype every third letter).

The screenshots on their site are a little small. And frankly after trying to take one myself, I realize why: text doesn’t shrink well. Anyway, I tried to take one from our last sprint planning meeting to give an example of what I’m talking about (click on it for the full sized version which looks a bit better):

Each person who logs into the gobby server chooses their own background color. Whenever a user types something, those changes are highlighted in that user’s color, which (in our case at least) is less about ownership and more about differentiating changes. Again, realize each user can simultaneously type in the same document, which leads to an incredibly collaborative environment. And yes, seeing a document come together in different colors like that just plain looks cool.

It’s been a while since I got a good picture of my 2 year old geeking out. My wife actually took this one after my daughter found my red fedora which was still packed from moving.

Then I found Conky. The easiest description is found directly on their site: “Conky is a free, light-weight system monitor for X, that displays any information on your desktop.” The addictive part is that there are hundreds of variables that can be displayed as part of its monitoring. Do I need, at all times, to see the amount of space left on my /home partition? No. Does the fact that I can do it make it awesome anyway? Hell yeah.

In the past week, I’ve had a number of coworkers see it running on my laptop and ask me what it was and how to configure it. The “what” part can be seen in the top right corner of my desktop below (click the image for the full size 1680×1050 version):

The “how”, or at least my settings, can be found here. I’m playing around with a personal github account for backing up and sharing this sort of stuff. I’m still not sure how much I’ll use it, but for now my conky config file can be found there. It’s a bit of a mess, but should be at least enough to give users an idea of what can be done. If anyone knows of any particularly cool use of it that I’m missing, definitely let me know.

I recently got into using Amazon for buying MP3s. It’s DRM-free and has a great integration with pandora.com, which lets me hear a new song and click directly in pandora’s interface to buy it. It’s technology at its finest, if not its most dangerous (to my credit card).

Amazon is annoying in the sense that you can download single MP3s as an MP3 file, but to buy a full album you have to use their proprietary downloader. This was bearable until a recent change where all MP3s now have to be downloaded using their downloader.

The problem? The most recent build of their client is for Fedora 9 (or Ubuntu 8.10 if you go that route). So their software is, um, I’ll just say “not up to date”. I was able to hack around to get it to run under Fedora 10, but it’s flat out busted in Fedora 11.

As I said at the outset, there is hope. Clamz is a command line app to download MP3s using Amazon’s .amz formatted files. It’s exactly what you think it is. You download the .amz file from Amazon (at the point in Amazon’s workflow where you should be just downloading the MP3 itself) and run the clamz executable passing in the .amz file. Poof, it just works.

I’m disappointed it has to come to this. I know Fedora 11 is still new, but that doesn’t change the fact that there was never a Fedora 10 build (and still is no Ubuntu 9.4 build) of the Amazon downloader. Serious good karma to the Clamz project for stepping up and filling this need (at least until I find a new outlet for buying MP3s; if Amazon doesn’t want to let me buy from them then I’m not gonna go nuts trying to).

There are many successful, thriving open source projects. TWiki is no longer one of them.The following two blog entries do a great job of describing the demise of the TWiki wiki platform (they do a great job, so I won’t even bother summarizing what went wrong).

TWiki is dead, long live TWiki!

Hostile Takeover of Open Source Project TWiki

Here is an excerpt:

“Trivial example here. The curricula mandates that you learn about data
structures and prescribes certain books and the like. However, the
actual implications of using various ‘like’ data structures or, an
appreciation of the trade-off between algorithms would come if you start
looking at code beyond what is provided by the syllabus.

Having a scientific temper – the ability to grasp the core of the
problem, figure out a pattern to search and adhere to some logical (and
sometimes illogical) reasoning is what will get you a job.”

As both a teacher and an open source developer, this e-mail really resonates with me. I began this semester of Algorithms & Data Structures II by attempting to describe the difference between a theoretical exploration of computer science and actual “real world” software development. I say “attempting” because I acknowledge that until you’ve been in both areas, it’s difficult to fully understand the distinction.

One of my biggest goals in my classes has been an attempt to expose students to both of these aspects. I make sure to cover the required material for the class. But I also supplement it with my experiences from working on real software projects. Being able to pull up one of the open source projects I work on and show students real applications of the concepts behind a domain model or where recursion would actually be used in a non-trivial example is an extremely valuable aspect of the class.

I recommend reading the entire contents of the original e-mail. It’s not too long and does a great job of explaining the benefits of real project work while pursuing an academic degree.

Read the full e-mail here.