While it is usually not hard to judge what is and what isn’t suitable material to publicly publish, it is considerably simpler to just privatize everything. On the other hand, I am certain that nothing I have blogged about thus far would reveal any company secrets or proprietary information. Thus perhaps I need to rethink my blogging strategy. I am considering trying to implement a “multi-blogging” system where I can feed blog posts to owenkellett.info, my internal Sun blog, or both. Although, I am not particularly fond of the software used by Sun (Roller) as it is buggy and not as robust or feature rich as WordPress. In any case, it is for posts like this one that such a system would be useful as it is suitable for both of my blogs:

As I’ve been becoming more comfortable with my responsibilities at Sun, I’ve also been growing more accustomed to the tools that they make use of. In particular, I am growing increasingly fond of Perforce, the version control system that we use. Previously, my only exposure to version control has exclusively been CVS, which I have used in a number of projects in the past. For the unititiated, version control software is a tool that coordinates the concurrent modification of a source code by multiple developers. Thus for example, if Joe Programmer and Cathy Hacker are both working on different segments of the same source code file, the version control system helps to cleanly integrate these modifications and all modifications made by a number of users at once. The general mechanism used to do this is by keeping a centralized repository of the source tree on a server, and each user checks out their own individual copies of this tree. Users than make changes on their individual trees, and check them back into the central server when they are finished. The checkin process controls the merging of individual files to keep the source tree clean.

There is no doubt that CVS is very powerful, featureful, (and also very free). I am no expert of either CVS or Perforce, but I have noticed a couple things about Perforce that make it “safer” (probably the most suitable word to use):

• I seem to distinctly remember a situation using CVS when I made changes to a particular file, and then perhaps stupidly did a “cvs update” before a “cvs commit.” The result was the version of the file in the repository clobbering all of my changes. Perforce forces you to explicitly “p4 edit” every file that you want to modify. Therefore if you sync up from the repository before submitting your changes, it won’t overwrite all of your work.
• Perforce submissions are stateful. When committing changes to a CVS repository, if something screwy happens to your connection or the server mid-update, you could be left with a bunked repository. This can’t happen with Perforce as “p4 submit” operations are atomic.

On the other hand, Perforce costs money. I really need a lot more exposure to both systems to make any real assessments as to which one I prefer. However, these are just a couple quick things that I noticed that I thought were worth mentioning.

In any case, even though I’m still very much a newbie, I’m starting to get slightly more involved at work. One of the tasks that is currently on my plate right now is setting up a system to completely automate the process of installing a number of different potential Linux distributions (i.e. Red Hat, Fedora, SUSE, etc.) from a remote server. The reason for doing this is to streamline the testing process for a new server system. Essentially, the design goal for this system is that it ultimately needs to officially support not only Solaris, but also several different versions of Linux as well as Windows. In order to automate the test process, we need to be able to sequentially roll out an installation of some OS, test that OS, and then move onto the next supported OS. All of this needs to be completely hands-off so that the actual testing process can be done any time (e.g. overnight) and without any user intervention.

A typical system that would be used to accomplish this task is a PXE server which meshes DHCP services for dynamic network discovery and configuration with TFTP services for hosting OS images for automatic download and installation. Now from my own personal research, the concept does not seem terribly difficult. One caveat, though, is that PXE is not often used for completely hands-off installations (i.e. users often must select menu options, etc.) and of course things that seem conceptually simple are always more complicated to implement. Additionally:

• Before a few days ago I had never heard of PXE.
• My boss does not have any experience setting up PXE services either. (Of course that’s why it’s my job to research and carry out an action plan)
• None of the people in the Automation group that I work have experience setting up PXE servers.

Now all of these facts simply make it a challenging task and force me to be resourceful. I have already met with one person from the IT department that has some experience with PXE servers and actually has one set up that we might be able to leverage for our ultimate goal. Of course working with him could prove to be a problem. Because he’s an “IT person,” that means he’s never in his office, has too many things to do, and is needed by 40 other people all at the same time. In any case, we shall see how it goes.

• I’ve only been there a few days and I am already confident that Sun will be a great environment to work in. Sun has no cubicles (I have my own office with a large work area, a door, and a private quiet space if I need it), casual dress (jeans to work every day), flexible hours (I’ll prob. end up taking 1.5-2 hrs for lunch every day so that I can workout/play frisbee), and most importantly very nice and very smart people (from what I gather and whom I’ve met so far).
• Commuting 40 minutes to an hour each way every day is tiring. I’ve worked plenty of 40 hour weeks before at O/K, but that was a 7 minute commute. Driving in traffic for that amount of time every day is definitely more taxing on your body.
• It’s tougher to come up with the energy to blog after working all day. Interestingly, Sun encourages their employees to blog, and provides their own blogging service for anyone working at Sun to use. I haven’t really gotten deep enough into any projects yet to be enthusiastic enough to write about them, but hopefully I will soon and start occasionally writing while at work. (I’ve also set up a super secret Sun blog that is internally accessible to Sun employees only. Any bloggable material that contains potentially overly revealing information will go there!)
• And perhaps the most overwhelming emotion that I’ve been experiencing while starting my first full time job: I really miss Katy. She’s meant so much to me and to my personal growth as a person over these last 3+ years. Talking on the phone is nice, but I really wish that she was here as I enter into the next chapter of my life.

Back to work tomorrow.

Unfortunately, though, the less than optimal sleeping conditions made for a terrible night’s sleep. We didn’t feel like carrying sleeping pads and were just sleeping on the bare wooden floors of the shelter. At around 2AM I woke up and heard something rummaging around in the trees just outside the shelter. I looked out and saw Brian roaming around collecting branches and wood to build a fire. Apparently he was so miserable trying to sleep that he decided to just get up and occupy his time by building a fire and keeping warm. After trying to sleep a little longer, I finally gave up and sat around the fire with him.

It was actually pretty nice to hang around the fire in the middle of the night like that, but it definitely caught up to us on Sunday. We stayed up until dawn and left basically as soon as the sun came up. We were both soon so overtired that we pretty much just wanted to get back to the car. Regardless, though, it was a good trip and a good way to spend the weekend.

In any case, I still have not taken the plunge and installed Vector, but there was one interesting discussion in the above mentioned review that caused me to take a parallel investigative leap. Up until now, I have relied on both KDE and GNOME to serve my Linux desktop environment needs. For the past couple of years, I used KDE on Mandrake up until just recently when I switched to GNOME on Ubuntu. In both of these cases, I was using my old PIII 900Mhz, 512MB Thinkpad T22. The problem that I’ve run into lately is that GNOME is beginning to have a clunky feel to it on this system. It is well known that GNOME is fairly memory intensive; combine this with an annoying Firefox memory leak, and you have yourself a system that can really bog down. I tend to carry out several parallel tasks using open applications on multiple virtual desktops. The result is a lot of hard drive thrashing because the system keeps running out of physical memory.

Coming back to the original Vector review that I mentioned above, an alternative desktop environment called Enlightenment is discussed. I have seen Enlightenment before, but apparently there is a complete rewrite going on and the newest version, E17, has reached a usable form. Intrigued, I decided to install it on my Ubuntu system and see what it’s like.

The first thing that I will say about E17 is that in its current form, it is very raw. Still in a pre-alpha stage of development, a lot of the functionality isn’t quite there yet and at this point, it is certainly not for the weak. However, it is in a stable form and most of the drawbacks of using it lie in the fact that many of the configuration tools are not very polished and a lot of command line tweaks need to be done and support files need to be created and modified. Once you get it set up, though, it’s a desktop experience like none other. As it is mentioned in the Vector review, once E17 is finished, it will bring Linux on the desktop to a whole other level. It of course has lots of eye candy, it’s light years ahead of most other GUI’s in terms of efficiency and functionality, and most importantly, it does all this with a small memory footprint and no need for any hardware graphics acceleration.

Have a look at a screenshot of my desktop that I started working on configuring today. I’m still trying to muck around with all of the currently available configuration options and modules, but even in its present state it works very well.

Click on the map below for a much larger view of our planned route. Also, if you want to join us let me know! (seriously)