Posts from — February 2008

I’m planning on going to New York City in a few weeks to attend a cryptography conference but while I’m there I plan on visiting Columbia University to investigate the distance-ed engineering program office for the Columbia Video Network.

I’ve mentioned it before but it’s looking more and more like a great option for me. The program is rigorous, bears no difference in coursework with the on-campus program, and can be completed completely off-campus. As I’ve come to this realization, I started to look deeper at the classes that I might sign up for.

That’s when I saw that some of the most notable people in the Internet’s history were teaching classes offered by CVN. One in particular bears mentioning: Steven Bellovin.

Prof. Belovin is co-teaching a class called COMS W4180: Network Security and it looks like it might be an excellent example of a class that offers a world-class professor to a wide Internet based audience. Prof. Bellovin had a front row seat for the development of the Internet and is probably at least partly responsible for many of the protocols and mechanisms that we use today. As a grad student, he invented USENET, probably the first and most widely used means of distributing information across the Internet even today. He also wrote the famous “Firewall” book published by O’Reilly that I used to understand and configure them. It’s going to be exciting having the opportunity to take his class. I’m certain that he can offer a lot of context to the problems of security on networks.

This brings up another point: you should figure out what you want to study in graduate school and go to the school where that is being studied. I live close to several world-class universities but only one of them is doing anything remotely connected to network security or cryptography. A couple of others are focusing on other topics like bioinformatics or wireless networks. Most of the rest don’t even have a computer science program of note. Having faculty support for a particular area of study is important and if you are all by yourself studying some field, the chances that you’ll waste your time on unproductive areas of research goes up tremendously.

Since I don’t have the luxury of attending a traditional program, I’m more constrained than others in the choices of programs that I can pursue. It’s a lucky break that there are programs out there that will cater to my choice of field but still permit me to maintain full-time employment.

I recently decided to install Linux on a Panasonic Toughbook CF-28.  In case you aren’t familiar with this particular piece of hardware, it’s essentially a laptop that thinks it’s a tank.  It’s heavy, slow, and super sturdy.  When you close the lid on this thing, it sounds like you’re slamming shut the door to a prison cell.  Supposedly, you can drop this thing on a concrete floor and it’ll still work just fine.  The laptop in a cop car is likely to be a Toughbook.   My particular model has a rubberized keyboard which will resist spills.  (Just the thing for that trip to the desert!)

Anyhow, my machine (obtained from eBay) did NOT have a CD-ROM drive, only a floppy.  This is quite a problem if you need to install software.  I could have returned to the eBay well once more but it seems that CF-28 CD-ROM drives are quite the prize.  Undeterred, I decided to try another avenue of installing software sans CD.

It turns out that Fedora 8 offers a package called cobbler.   With this package, you can support clients that boot over the network.  Since Fedora doesn’t offer floppy installs anymore, I tried this route.  I set up one of my local machines to serve as the bootserver that would respond to requests for netboot images and had at it.

Fast forward several hours of making PXE bootdisks, importing a Fedora 8 DVD distribution and then going through the whole install process with the Toughbook and I had a working install of Fedora 8.  Except that X would not start.  Since one of the applications I needed positively relied on X to work, I needed to fix this problem.  I tried system-config-display, hand editting the xorg.conf file, and then going through the xorg.0.log to find the problem.   Nothing worked.  I decided to try another approach.

I pulled the drive from the laptop containing the Fedora 8 install and installed a blank drive.  Then I visited the debian.org site, downloaded the floppy boot images, and used them to bootstrap an install using a repository on the Internet for the files.  About an hour later, I had Debian installed on the machine, X was configured and working properly, and it even gave me the option to set up the drive with a LUKS encrypted partition.  Easy.  So Debian it is.

Of course, the big question is why Fedora (which uses the X.org distribution as well) didn’t detect and set up the configuration as well as Debian did.  I could always just copy the xorg.conf from the Debian install to the Fedora one and see if it all works.  In the end, I might decide to do that.  But why?  So far I have a working machine with a high security encrypted drive as well.  Why mess with a working platform?

Hats off to the Debian folks.  You got it right on the first try.  No muss, no fuss.  Just a solid distribution.

One of the better explanations of the P vs. NP problem.

I’ve been looking at the whole Ruby on Rails thing as the foundation for a kind of web application I was thinking of.  Right now, it’s purely a personal thing.  I’m doubtful that it has any commercial potential.  Nevertheless, it’s still a good way to get familiar with the whole process of getting the right libraries and compilers installed.  Its even more interesting when you try and use some of the more modern development tools for managing a project.

I’ve been playing around with Ruby on Rails, Eclipse, and Subversion lately.  I’m planning on posting a HOWTO in case anyone is interested.  As it turns out, getting a MacBook Pro set up as a development platform using some of the more interesting tools isn’t as straightforward as it seems.

Stay tuned.