Scenario:
kontact (mail app) on Desktop 1. Surfing for a new car on Desktop 3 . New mail comes in, switch to D1, it has a link, I click. Natively, firefox opens the link in my D3 window, ruining my task-oriented desktops (and making me switch to D3 just to see the opened link). What I would have liked is a new window of firefox to be opened in D1, at which point I can decide to move it to another desktop (extremely quickly with my bound keyboard shortcuts), or if it’s a quick task, do it on D1 and be done with it.

This depends on wmctrl (debian/ubuntu: sudo apt-get install wmctrl).

Put this file somewhere (I chose /usr/local/bin/firefox-wrapper.sh) and tell your WM to have it as the default web browser (for KDE: kcmshell defaultapplication).

firefox-wrapper.sh

I put together a quick little script that’ll poll the proc filesystem and toggle the SPDIF out accordingly.

Check it out: MacBook LED

After screwing around with the ALSA driver source code, the Windows driver files, and the Microsoft and Intel spec sheets, I’ve found a solution that allows for both the microphone and the headphone sensing to work! Read more in the Linux on MacBook section

They support many platforms (at least Windows, Linux, and OS X), but one problem I’ve found on the *NIX is there isn’t a way to translate from the nicknames of those in your room to IP addresses without having to refer to the list. I’ve created a bash script to parse the list of Hamachi users in your rooms and place them into your hosts file (/etc/hosts typically).

The general naming convention I use for my Hamachi nicknames are [computer owner’s name]@[computer name], or sometimes even just the [computer name]. An example of the former nickname style would be jason@mooBook (where mooBook is my laptop’s name), and the latter mooMedia (where mooMedia is my media server–like the moo theme?). So, my script will take the [computer name], tack on the room name, and map that newly created hostname to the VPN IP of the machine. Suppose both of these machines are in the room mooNet. I can now ping moomedia.moonet (case insensitive) or moobook.moonet. I’ve tacked on the room name to prevent adversarial behavior (although if you’re in the room com or net or any other TLD, you’re screwed ;).

You only need to run hamachi2hosts.sh once initially and when more users join your rooms. It creates a section in your /etc/hosts for the automagically mapped Hamachi hosts. Please be sure to change the settings near the top of the file. Also, it assumes you’re logged in when you run it.

Here you go: hamachi2hosts.sh

We waited in line for Wii Sports (the packaged game with 7 different sports), and it was pretty darn cool. I played tennis and baseball, both which were fun at first, but could lose their appeal after constant playing. BUT, the concept is amazing, and from what I could tell, the hardware definitely has capability to capture precise movements (Wii Sports was overly simplified, I couldn’t throw a ball even if I tried. I still got my ass handed to me in tennis, heh). The controller felt pretty natural, and the graphics weren’t that bad from what I could tell (and taking the price into account).

I hardly ever think about buying brand new consoles, but the Wii is REALLY tempting.

What this lets you do is have a single XP installation, that is able to be booted natively (for games, etc.) and in the virtual machine (from Linux for random things, mostly turns out to be opening Office files that don’t load cleanly in OO).

I’ve updated the Linux on MacBook’s VMware section with detailed instructions on how to achieve this. On the MacBook, I had issues with XP BSODing on boot, so I’ve noted how to fix this.

sudo sh -c "echo 3000 > /sys/devices/platform/applesmc/fan0_minimum_speed"

Enjoy a cooler laptop!

kfmclient exec FILENAME

So, I’ve added an alias to my ~/.bashrc:

alias kopen='kfmclient exec'

and can conviently type:

kopen README.rtf

to open README.rtf with the bound handler.

The first issue I had was the mouse pointer jumping a few pixels constantly when moving around. The trackpad consists of many sensors (20 on the X axis, 10 on the Y axis for the MacBooks), so when you glide your finger across, each sensor on your finger’s path will go high as the finger approaches/hovers (and go low once your finger leaves the sensor’s ‘zone’). For the trackpad to be used for neat functions (two/three finger functionality), the synaptics X driver should be used, and in order for it to be used, the trackpad must give off absolute coordinates instead of relative (so the actual position of your finger(s) as opposed to the change in your finger(s)’s position). There is thresholding for each of the sensors in order to discard jitter when the finger isn’t triggering a sensor. As soon as a sensor passes the threshold, its value is used (along with the other sensors that are over the threshold) to calculate the absolute coordinates. So, when a finger is approaching a sensor, the sensor’s contribution to the absolute coordinate calculation is 0, but when the finger gets close enough to allow the sensor to pass the threshold, the sensor’s contribution becomes at minimum the threshold value. The default threshold level is 10, so this causes the sensor to go from giving 0 feedback on coordinates, to giving a value of 10 (not exactly) to the function to give coordinates. So instead, I subtract the threshold from each sensor that contributes to the absolute coordinate function. This ends up giving a smooth mouse movement when a sensor suddenly goes from below the threshold, to above.

The second issue was the accuracy of two finger (and three finger) detection. These are used for two-finger scrolling, two-finger clicks, and three-finger clicks, so it’s pretty important in day-to-day use. I found it difficult to trigger the multiple fingers properly (I couldn’t have my two fingers touching each other and I would have to make sure they maintained a good gap when scrolling), so two-finger scrolling was difficult to do. The way it checked for multiple fingers is to count how many sensors that has passed the threshold are preceded by a sensor that is below the threshold. So, the first finger would obviously pass, and then if there is a sensor gap between the first and second finger, the second finger would get triggered. I didn’t like this as I don’t usually leave a gap between my two fingers (OS X didn’t have this constraint). I changed it to check for ‘humps’ in the sensors, so basically check for transitions from nonincreasing to increasing sequential sensor readings on each X/Y axis. This works since the middle sensor(s) (sensors are close enough so when having two fingers down, there will be one sensor between the two) can still be over the threshold and still count as the lower part between humps. After the patch, the two-finger scrolling works as well as OS X (from what I can tell).

The last small issue is being able to change the threshold value. I turned this into a kernel module option (modinfo appletouch).

This patches against Mactel appletouch patch that allows MacBooks to work with that driver. I’ve submitted it to the maintainer of the appletouch kernel driver, so we’ll hopefully see it there!

appletouch-2finger_det-threshold_opt.patch