gir

(no subject)

i really hate trying to track down patent licensing terms. by comparison, i've come to realize that i merely dislike software licensing.

once upon a time, i was a software engineer / researcher type. i miss those days.
gir

cards

if you would like a card, plz 2 comment (screened) with address info. next batch will go out shortly.
gir

(no subject)

with ohio called, all i can say is: congratulations, president obama.

what is left to see is to what degree obama has a mandate / landslide, and whether the culture wars are now, finally, over.
gir

(no subject)

long time no see, i know. that should make this posting all the more special.

i voted in my first election. if you are registered, go out and vote. i don't want to hear any bitching and moaning about it.
gir

(no subject)

it's been a while.


in no particular order:



  • aunt's 60th birthday was last week, so i flew out to detroit. ended up playing bartender since the actual bartender showed up smashed. gift of a digital picture frame went over really well. i really love my garmin --- it made getting around michigan (which, i'm beginning to realize, is as impossible to navigate as virginia because it's just as flat, just as green, and just as generally lacking obvious landmarks) a ton easier.

  • the same tactics used to claim that intelligent design is "science" are being used to claim that global warming is a hoax. it took all my strength to not yell at the people across the aisle from me on the airplane when i happened to overhear them discussing how they thought global warming was a myth. turns out that the right has, in general, been monumentally successful at getting people to live in a world that simply doesn't correspond with the reality i experience.

  • finished stage 1 of computer build, which involved adding more ram and a dlink pci-e wireless nic. in the process, i discovered that comcast botched the cable installation on my apartment (quelle suprise!) and didn't turn on cable in the guest bedroom. not a happy camper. i think that i'm getting rid of them pretty soon, maybe altogether if quest has reasonable prices for naked dsl.

  • work has been meh. they lost about half the company's health insurance stuff, which means that it took about six weeks extra for us to get our cards. that's annoying, but what really galled me was the reaction when i went to ask brave and fearless leader if (a) he knew about the situation and (b) if there was maybe something he could do to help. i've been spending a lot of time on the phone lately trying to work with the support desk to get things that had been working working again. it took me a while to figure out that axis network cameras have some really strange behavior. they'll auto-configure via dhcp, but they won't honor their dhcp leases expiring. i did finally manage to get a static ip for the thing, so the past two days haven't been a complete wash. actually, i managed to get a lot done w/r/t getting nighttime in good working condition. it's remarkable how changing a single, fairly simple assumption can totally break your software design (or, in this case, lack of design).

  • in other work news, i'm not really sure how happy i am about my title. i'm still trying to figure out how big of a deal that's going to be in terms of career mobility, but i'm more irked by the fact that when i read my current job description and compare it to what i had been responsible for prior to the merger, i keep on coming to the conclusion that i got stiffed. i suspect that they brought everyone across at the lowest possible grade that fit their salary and experience. the jury's still out on why tho.

  • most recent attempt to date resulted in epic fail. nice enough guy, but jesus... i'm not impressed by people wearing tri-force necklaces, zelda jackets, and bsg t-shirts. people who have nothing nice to say about their friends also don't impress me. not saying "thank you" or even acknowledging that i'd picked up the tab doesn't just not impress me, it pisses me off.

  • did pretty well in my controls class last semester — certainly far better than i'd expected to. the big lesson for me at least boils down to: arbitrary pole / mode placement doesn't buy you a damn thing if you don't know what those poles and modes should be. optimal control ftw! i'm about 80% certain that next semester will be stochastic processes, assuming that me being late to work 3 days a week isn't going to be a problem. if it is, i'm going to have to backtrack.

  • a couple conferences coming up that i'd like to go to. we'll see if i can get work to pick up the tab.


been looking at buying a condo, with mediocre levels of success so far. more on that when i get time.

gir

(no subject)

i just bought $crazy/2 worth of new parts from newegg today. there's a rational part of me that knows that a machine made with parts that are 5-8 years old on average simply can't handle capturing hdtv signals, but a little part of me weeps for its passing. that machine was my first "real" beast -- 2 x 2.4ghz p4 xeon with a whopping 1.5gb ram. it served me well through round 1 of grad school. here's hoping to an equally admirable performance through round 2.

i will purchase the remaining $crazy/2 worth of components once i get reimbursed for classes. then i will probably need to bite the bullet and figure out if it's even possible to do gigabit throughout the apartment. my next place will so be wired with a bigass switch in the basement.
gir

(no subject)

i am such a sentimental dork it hurts. seriously, the fact that i was swooning after watching (an admittedly incredibly romantic) scene where the lead arranged a special two minute date with his dermatologist.
gir

(no subject)

in no particular order:
  • it's taken me several days to figure out how boost's statechart stuff works. there's a part of me that's impressed with how it works, and there's a part of me that's kinda disappointed. three days of digesting documentation later, i have an approximate idea of how to make it do what i want. </p>

    the problem boils down to this: boost's statechart stuff gives a mechanism for encoding a statemachine in a rather clever, very strongly typed way. you can get notifications that you entered a state (by a constructor call) and that you've exited the state (via a destructor call), but there's no obvious way to say "while i'm in this state, perform these actions in a loop". i mostly grok why this is the case — they want statechart to do one thing well — but i'm a little disappointed because it seems like pretty obvious capability to have left out.

    there's a reasonably simple way to make this work: create an interface that exposes whatever state dependent services you need. make each state implement the interface in addition to all the statechart stuff. then do something like:

    while(...) {
    StateFunctor& lFunctor = reinterpret_cast<functor&>(lStateMachine.get_current_state()); lFunctor(lEventQueue); if( !lEventQueue.empty() ) {
    lStateMachine.process_event( lEventQueue.front() ); lEventQueue.pop();
    }
    }
  • there's a guy at work who likes to come ask my office-mate and i for advice or opinion on design questions / etc. he likes to ask questions, and when you give him an answer that he doesn't like it's pretty clear that he doesn't like the answer. three times this week, he's come back to tell us that after several hours of fucking around trying every other solution than the ones that dale and i have given him, finally, he's implemented one of the solutions we gave him. i'm taking wagers on how long it's going to take for him to realize that he needs to just shut up and do what we tell him to.
  • in other betting news, i'm pretty certain that the new "busdev" guy at work (he's officially a "product development manager") is our beloved leader's replacement. now, whether or not beloved leader recognizes that fact is a totally different story. either way, i've put in a PO for popcorn, stadium seating, and tickets so that i can charge admissions for the fireworks that will shortly be following.
  • i feel kinda silly. it's taken me 3 weeks to figure out that i need an open loop state space realization to compute feedback gains. i applied pretty drastic state reduction to my original system since it wasn't controllable. now it's controllable and observable and all is right with the world again. bonus: since i want to force all the system outputs to zero (the outputs are essentially errors), some things become a little simpler to do. unfortunately, it's a MIMO system with inputs coupled by the B matrix. unfortunately, this means that i can't just use the single-input stuff :-/
  • on that same theme, i feel like i've drawn block diagrams for and worked out the closed loop representations of just about every conceivable coupling one or two systems with state feedback, state feedback with integral action, state feedback with an observer, etc. i'd give even money that in doing so, i've solved at least half the problems that are going to be on the midterm.
gir

(no subject)

more python adventures: it turns out that a sane editor makes the white space nazism less problematic. if it weren't for the silly amounts of time i've spent reading api docs (btw, scipy: your documentation sucks), i'd actually seriously consider python for more actual work projects.



as it is, i'm working on embedding a python interpreter into a c++ program, partly as a way of testing an idea, and partly because it's about the most stimulating thing i've had to think about in quite a while (at least in terms of work stuff).



i've spent a lot of time working through state space stuff and doing a ton of blackboard algebra. i'm finally beginning to put all the bits and pieces that i had been missing into place. i should have spent more time this weekend doing more homework (and i will, after supper), but i'm feeling better about where i am with that class in general. i need to reread my notes and get studied up for my midterm which is coming up real soon now (tm).



spent a large chunk of today doing filing. i still don't quite believe how much crap gets sent to me. on the plus side, my old shredder died, so i picked up a new industrial strength one that swallows credit card spam whole. i am deeply pleased by this.