Recent blog entries for aaronv

Thanks, everyone, for certifying me ;). I promise not to cause too much trouble.

I wasn't able to finish the last step of the project yesterday. bah. Finished getting it to build, and it seemed to load right into my powerPC simulator, but I didn't have time to run it through and examine the tracer's output.

Oh well, on a side note my boss said he's giving me a $3 raise when I finish it off. That should be within an hour of starting next saturday ;)

I've never had a boss give me a raise before, ever. This job is a keeper. I mean, he paid me the first third of the summer for educational reading. get that! Even if I wanted to move on, I couldn't in good conscience; I feel that I owe him a little more work to make that worth his while.

There's a trick to flying. In fact, there's a knack. The knack lies in the ability to fling yourself downward, and miss the ground...

I've been trying that for weeks, doesn't seem to be working.

In the works, unofficial omniorb 3.0.1 debs. These are officialy brent's turf but I haven't heard from him in a while ;). So they remain unofficial until woody freezes. If we haven't seen anything by then I'll consider asking for permission to NMU.

I'm going to be done with my project at work today! hoorah! PPC assembly is fun, but it's getting old. Time to finish it up and devote more time to the Beowulf project. Always on the move ;).

Automating the process of cloning a cluster has been a pain in the butt. I've got it mostly working, enough to expedite the process for me (partition the hard drive, download tarballs, unarchive, boom! fresh clone) but the little finishing tasks are a bit too interactive to make it worth automating them. Problem is, it has to be 100% easy for someone who's not a unix god. Bah. That's the next one I have to finish up, shouldn't really take all *that* long.

We'll see.

In the meantime Farewell, and Thanks for all the Fish >:^P.

23 Sep 2000 (updated 23 Sep 2000 at 18:17 UTC) »
At an improbability of 2^3498 and falling...

Who knows how I ended up where I am, but it's great. School is great, work is great... Now I just need some more time for my comp sci hobby, god knows it's scarce ;)

classes have begun. School is great. I love it. I could go to school forever. Oh well, we all have to make a living ;). I'm taking:

  • ME 211 (statics for engineering majors) -- 3 units
  • EE 112 (introduction to analog circuit design) -- 2 units
  • MATH 242 (First order differential equations) -- 4 units
  • PHYS 133 (General physics, this focuses on dynamics) -- 4 units
  • PSYCH 202 (fluff psychology class, gets me a few units) -- 3 units

All around, my classes are fun. I love math and physics, and my major. I can't decide on a direction for grad school yet (whether it be a study of biology and interfacing biological/electrical systems, physics and CSC for quantum computing, or any number of other fields that fascinate me) but I do know that I want it to have something to do with computer science.

So what do you think? I'd like to hear from people in hybrid fields. I really think that's the only way I could go; I don't want to be just a 'computer engineer' or a 'computer scientist' or an 'electrical engineer'. I also want to be an academic the rest of my life, but I need to work on my study habits if that's ever going to happen ;)

Anyway, at work I've been reading every article that comes through about CFD and other types of physical modeling for error-bounded simulations. Some of them from mike warren himself, a guy who my boss is collaborating with on this project of ours. All of my classes (except PSY) apply perfectly. I've been playing with a little point charge/mass simulation code myself...

And all I can say is.... MPI is amazingly elegant! Props to the standards committee.

Stairway to heaven

It's quite ironic. I spent much of my time last year playing with Berlin, preaching the coming of the GUI developers' messiah.... Skipping classes to catch up on my Debian E-Mail... Generally being a groupie. A newbie groupie at that. I can't honestly say that I ever truly made a significant contribution to either project, but both combined have helped me become what I am today: a unixhead in the software industry.

I and a friend of mine, Dave, always share any kind of pure-science-related knowledge we happen upon; both of us love anything from any field (math, chem, physics, comp sci) that stimulates thinking. You would often find I and Dave in the study lounge at 4 AM the morning before a big test, sitting behind our books as though to study, but rambling on about various algorithms, new emerging fields of science, physical phenomena, philosophy, you name it..

Well, Dave turned out to be interested in computer science (he's a physics major; funny, I'm a compE major interested in physics) so I began preaching the savior-of-the-world, CORBA, to him, and drew him some ugly flowcharts of what happens inside Berlin, to the best of my knowledge at the time (considering I have never really taken enough time to grok any of graydon's/stefan's code to my own satisfaction). The idea sounds really great, especially on paper...

Well, he hit me with a hard question. I had a difficult time explaining CORBA (the telltale of deficient understanding), so I fell back on how Berlin eliminates dependencies on various widget APIs with a stub library... That he was contented with. Then I proceeded to explain common paradigms (pixel-oriented displays) vs fresco-style vector graphics. After this I showed him a cool demonstration of the first DR of Berlin's second incarnation: three buttons that, however you transform them, just magically intercept events.

His question was: why? Doesn't X do translations relatively well? Who cares if you can zoom/rotate/skew an application's image, embed one within another, et al if the cycle cost is so high? I couldn't think of an answer at the time; that's when I decided I was being a groupie.

I've since figured things out a little better.

Since first quarter, I've been relatively inactive on lists, and even more inactive in the goings-on; either get involved or stop wasting time, I told myself. Dave and I kept discussing anything and everything when we should have been studying (He still got A's; it was never quite so easy for me). Quantum computing, physical simulations, et al.

[time warp] a few weeks before the end of the school year, a professor of mine contacted me with a job offer. I'm working for him now, being paid to administer servers, construct a cluster (just one for now, this is a startup; more later), learn MPI and study just about any new scalable parallel application. I love this field, and what's even better we're going to be doing lots of algorithm research in the field of physics. Dave is signing on to help (as a new brain to pick at), and I believe that much of our code will be open.

For anyone who has read this far, what I'm trying to do is convey both my thanks and my apology. My thanks for putting up with me and infusing me with interest in these fields, answering my questions when I had them. My apologies for leeching, scantly returning the favor. You will hear from me on some kind of basis in both debian and Berlin, but I am not sure of my future participation; everything depends on time.

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!