P. Oscar Boykin

I received my IBM T42 laptop. Debian GNU/Linux runs well on this machine. I found pages by William Stein and Sanjiv Sinha particularly useful.

One issue I did have that I did not see covered by them is a problem with my Palm Tungsten T3. Using the USB sync along with the USB 2.0 controller (ehci_hcd in linux 2.6), I get no joy. My work around is to only load the USB 1 controller (uhci_hcd) and not load the ehci_hcd. It seems I am not the only one to see usbserial problems. I hope that they are all fixed before I need to connect a USB 2.0 device.

The only thing that seemed to cause me some small frustration is the external display. Finally, the solution I settled on was the following. Add this line to my XF86Config-4 Device section:
Option “MonitorLayout” “LVDS,CRT”
and make sure to boot WITHOUT the external monitor connected. By default it seems to send the same output to both displays at the same resolution. If the projector cannot handle 1400×1050, I switch to 1024×768 using the Gnome desktop preferences screen resolution applet.

Finally, I did use the IPW2200 driver for my wifi. I ran it by installing the kernel-headers package for the kernel I am using, and compiling the ipw2200 driver by telling it the directory to my kernel headers package. I did need to manually modprobe crc32 if I wanted the wep modules to load properly (but since I use a VPN and ssh, I don’t need wep).

Update: 9/27/2005
I found this mailing list post which explains my ehci_hcd problem with the palm. Actually, if I connect it directly to the laptop and not through the dock (which has a hub) everything works fine. So this appears to be a bug with the way some full speed devices interact with high speed hubs.

I took a few minutes today to finally print forms to register to vote. The EFF has a voter registration page. I tried to find forms locally, but no luck. I will get these in the mail tomorrow.

Here in Florida, the deadline to register is 29 days prior to the election. Today is 46 days prior (just 17 days left to register).

Slashdot has found the percolation search result.

It is amazing how little the people who comment bother to read. This work is really about showing that global P2P searches can be scalable and giving such a scalable algorithm.

I like the conspiracy theory comment. If this guy is correct, Google must have failed to get in touch with me. Where are my Google stock options?

It seems the percolation search is getting some attention. It won the best paper award at IEEE P2P2004. TRN Magazine is running a story on our percolation search algorithm.

The algorithm itself is really simple and could greatly improve systems like Gnutella and Kazaa immediately, but it also has more applications. Percolation theory is a really fun topic. In this paper, it is exciting to see this theory harnessed to greatly improve scaling of distributed search.