I just came across the Reia programming language, which is a language designed to run on Erlang's BEAM virtual machine. Another language I have recently been toying with is Clojure, which is a really nice Lisp implemented on the JVM. Reia, Erlang, and Clojure are all deliberately and carefully designed with language support for high-level, lock-free concurrency abstractions.

I'll come back to Reia in a second, after a long-ass detour, because it's actually Reia that knocked this bit of head dirt loose, not Clojure or Erlang.

Immutable/persistent data structures and concurrency

Clojure's secret sauce is it's data structures, which are specifically targeted towards concurrency. The data structures are immutable and persistent. The word "persistent" in this context means that when you "modify" it, you don't really alter it, but instead create a new value that incorporates the change. (As opposed to the other meaning of being written to disks, databases, punch cards, or whatever the kids use these days.)

This means that, conceptually, every time you change a persistent data structure, you get a whole new, slightly tweaked version. Conceptually. The important bit here is that because the data structure is also immutable, you don't really have to copy the whole thing, your implementation can share in-memory structure with the old version behind the scenes and no one will be the wiser.

Immutability also means you don't have to use locks to protect shared data from concurrent writes by multiple threads. There are no concurrent writes to protect, because there's no such thing as writing into an existing data structure. When you "update" some data in Clojure, you make a new "copy" of it with the change you wanted, and concurrent readers never notice because the original is still intact.

The single-assignment spectacle

Now, Erlang is heavily influenced by Prolog and has single assignment variables. What this means is that once you bind a name to a value, you cannot change that binding, the name is forever (till the end of its scope) bound to that value. Essentially, name bindings are immutable. The data structures in Erlang, like Clojure, also happen to be immutable, so there's no getting around this with funny business like binding a name to a data structure and then mutating the data structure.

Reia is designed to run on BEAM, the platform of Erlang. Reia does not have single assignment variables. Curiously, this is a very controversial feature of the language, with many people accusing it of breaking the entire concurrency model of its host platform. I found it extremely odd that it would be so controversial, and haven't really seen any good arguments against it.

Clojure does not have single assignment either, and gets along just fine. The reason it's such a non-issue is manifold. Most fundamentally, bindings are thread-local in all three languages: it is simply not possible to have a data race as a result of a mutated binding. Clojure is unique among the three in that it does have non-thread-local names, but they are explicitly declared as such and have extra concurrency semantics that are enforced by the language, again preventing data races.

Additionally, data races caused by mutating data structures via aliased references are not practically possible in any of these languages, because data structures cannot be mutated at all, and (in Clojure's case) shared references must be mutated under language-enforced conditions. (By "practically" I mean "using the subset of the language not including the shoot-yourself-in-the-foot library backdoors.")

Self shadowing, not mutation

One of the more thoughtful objections to Reia's mutable bindings I've seen was the idea that mutations to a name binding could be observed by another process if you closed a fun over the name and sent it to another process. This is not actually the case as far as I can tell.

Reia's conceptual model seems to be more accurately thought of as allowing bound names to shadow themselves. All the code after each assignment operation is actually seeing a new environment where the name has a new binding that shadows the old one. In this way, when you close over a name, it captures the most recent environment in which the name was bound. Subsequent rebindings are irrelevant because they don't really mutate the binding, they shadow it. Because Reia's data structures (presumably being the same as Erlang's) are immutable, the mutable structure funny business previously described cannot be accomplished either.

In a manner of speaking, Reia has persistent bindings.

Why single-assignment then?

Conflating single assignment and concurrency in Erlang seems to be a pretty common thing, so perhaps I'll suggest a better theory about why Erlang is single-assignment: Pattern matching.

Erlang's "assignment" operations (both with equals and in implicit positions) are not assignments at all, but pattern matches with destructuring bind. You can use bound variable names as part of a pattern in Erlang, and the match will fail if the value of the bound variable and it's counterpart in the datum being matched against don't agree:

  Foo = 42,
  [Foo,Bar] = [42,0],  % Success, binds Bar = 0
  [Foo,Bas] = [40,0],  % Fail, Foo does not equal 40

If Erlang did not have single assignment, the last two expressions would be ambiguous: does it mean rebind Foo in a destructuring bind? Or should it use its value as a pattern in a pattern match? With single assignment, there is no rebinding, so it is always unambiguous: unbound variables get bound in a successful match, bound variables are used as part of the pattern.

Reia disambiguates this differently. It uses a "splat" operation which can be likened to "unquote" in Lisp. This is best described with an example:

  foo = 42
  [foo,bar]  = [40,0]  # Destructuring bind, rebinds foo and bar.
  [*foo,bar] = [40,2]  # Successful pattern match on foo, rebinds bar,
  [40,bar]   = [40,2]  #   evaluates as if you wrote this.
  [*foo,bar] = [42,4]  # Failed pattern match.

The second line shows a plain old destructuring bind. All variables in the pattern are considered L-values and rebound. The third line demonstrates the splat, which can be thought of as unquoting, or causing foo to be evaluated as an R-value. Thus it is the same as writing the fourth line, where foo's value is substituted in the pattern. And finally, the fifth line would cause a failed match, again it does not rebind foo.

I'm pretty sure this theory behind single assignment has a lot more traction than the "it has something to do with concurrency model" theory that I've not seen anyone fully unfurl yet.

Anyway this was a long-ass post, sorry about that. I had to at least attempt to correct the crappy FUD about Reia's rebindable names. I'm actually really excited that more languages are blossoming top of BEAM, it really is a remarkable platform. I always wondered how hard it would be to write a BEAM backend for PyPy...

Well that was a freaking awesome weekend. Bikers URRRRRRVERYWHERE! I could have done without standing in 60 MPH stinging mist on my face for about half of Sunday, but I am not so easily defeated when it's the first thing resembling vacation I've had since .... November?

The 125cc started in fair weather but was red-flagged towards the end as Ike careened thru Indiana. It rained thru most of the slot for the 250cc race, which they initially rescheduled until after the MotoGP race. Finally the rain headed out and the MotoGP guys got to run, but the wind started kicking up about 10 laps to go. When that awful mist began, they red-flagged it with about 8 laps to go. After the wind refused to die down, and the Yamaha tent blew down and (I hear) hurt a couple of people, they eventually cancelled the 250cc race altogether. Boo :(

But hey, the weather was frickin awesome on Saturday in spite of all forecasts. I ended up getting about a thousand photos, which is decent I guess. I was expecting to take more, but ended up that I spent most of Sunday trying to keep rain off my front element. Mayhap I will post some more pictars soon.

The drive back south was sort of surreal. Pretty much the whole trip back was dotted with wind damage to trees and light structures. We stopped in Jeffersonville to try and find something to eat, and could not figure out why A) nobody was at Hooters at 7pm on a Sunday, and B) why there were 8 quadrillion people standing around Applebee's. Eventually it occurred to us that the power was out to most of the places (excepting Applebee's) on the drag we chose.

We ended up at Za's in Louisville proper, to find it packed with people, and only two waitresses. Turns out the entire highlands had massive tree damage and power outages; I guessed most of those people lived around there and had nothing else to do. It wasn't until our way back out that we started seeing the telephone poles snapped off in the middle and wires everywhere, and that most of the side streets were completely pitch black. Very strange to come back to that not really knowing what had happened.

So, all in all, a very eventful and fun weekend! What I want to see next year is a few more people in our group! "Wahh! The end of August is SOOO BUSY!" .. *rolls eyes* :)

So I'm here with cbrus at the inaugural MotoGP race at Indianapolis Motor Speedway. Yesterday was practice and qualifying for 125/250/MotoGP, and the AMA Rookie's Cup. I rented a EF 70-200 F/2.8L IS USM lens for my DSLR, and man it's making me look good. Sally forth for a few selections!( Read more... )

I often see people getting into debates about computer performance, and I find it fascinating the amount of cargo-culting that goes on in the name of performance.

I was reading this blog post by Jeff Atwood, wherein he recommends (among many other things) having as much as 8GB of physical RAM for a desktop machine. Jeff doesn't give any reason for recommending this much, and indeed unless you're using 64-bit build of Windows (more like "enduring", with all it's attendant driver bogosity), all but the first 3GB is ignored.

But then I saw this comment, quoted in part:

And, in my experience, all that work [of fussing with 64-bit Windows] doesn't really get repaid much, as I've never exceeded more than 3 GB of physical memory use.

What follows is mostly verbatim from a comment I left on Jeff's blog responding to that. However, I find myself writing such epic comments often enough that I thought, "I have a blog of my own that mostly langushes. I'll put them there."

( So, onward... )

Just cleaning some comment spam, so here's a joke:

One DJ says to the other, "You want to go to a movie tonight?"
The other one says, "I dunno, who's the projectionist?"

Indeed I have installed Windows on one of my machines, for the purposes of running my particular brand of video editing goodness, Avid.

Using Unix more or less exclusively for such a long time certainly affects your outlook on what a computer should do. And how.

I wanted to share my external disks with the Windows box, so I had to install a Windows Ext2 filesystem driver. The thought of partitioning a 120GB and larger disk with FAT32 makes me physically ill. I will opine that it is high time a new lowest-common-denominator filesystem was adopted: with the size of today's disks, FAT32 is becoming painfully overextended. I was disappointed, but not surprised in the least, that my friend's PS3 would not mount my handy-dandy Ext2-formatted super-mega-huge external disk. As disturbingly and uncharacteristically friendly and cooperative as Sony decided to make the PS3, I suppose it *is* still Sony. Baby steps, and all that.

Back on the Windowsified box, I endowed it with a respectable text editor. I picked up Cream because I'm a Vim man, and Cream is a nice layer of Windows conformity for Vim, but leaves all my favorite obtuse yet ridiculously useful Vim commands, at which Notepad pilots shake their heads in solemn confusion and pity, within easy reach.

Next, I had to install MSYS because I quickly found Windows's command prompt damnably impotent. I know, "What the hell?" you say. I guess dropping to the command prompt when the GUI foils me has just become second nature. But Microsoft's token implementation of readline-like features behaves differently enough from a "real" shell that I find most of my muscle-memory from unix shells ends up biting me in the ass on Windows. It's more irritating than not having readline at all. Being able to hop over to that loyal dollar sign really soothes the withdrawal.

I also found having a copy of 'make' around certainly makes life easier when using Aqsis (Renderman). I wish I had a nickel for the number of times (prior to automating with make) that I edited a shader and rendered my scene 3 or 4 times in a row, scratched my head at the lack of effect, only to realize I forgot to recompile my shader. And then got pissed because the command I was using to compile the shader wasn't where I expected it to be in CMD's folded-ass up-arrow history. MSYS to the rescue: make, enter. Whee!

I have thus far managed to avoid the temptation to go batshit insane and install Cygwin or coLinux or something. We'll see...

In typical ADD form, I've noticed like at least 100 times now that all the links on here to my "stuff" that I was hosting on my server are broken since it's IP is changed, and every time I notice, I think "damn it, I need to fix that."

And then I promptly forget about it.

I bought a domain name for my site (and I even got a domain hack :D ) and it's all glued up to a super-suave Xen VPS, so hopefully this problem won't happen again.

The only problem is, the person I'm most often sending links to my stuff to, can't get to the server from his home internet link. The traceroute ends somewhere in Verizon's infrastructure, in some asshole box full of microchips in Richmond who would like us to believe it is a router.

*fizz..*

I sure hope I haven't lost The Knack.

It's been 37 weeks. Sheesh. I was working on my website, and got nearly everything done except the front page. Those are always the hardest part for me. I'd have to say that's probably why the LJ has languished for this long. I'll get it done, some day!

The weekend plans went off without a hitch, which is welcome relief from the typical outcomes of my big plans. Thanks to rocketpup and joeypan for the best weekend in recent memory. I'm still grinning like an idiot.


Me: i don't know what the hell to write
cbrus: hmmm.. write "i am a pig"


Well apparently, "I am a pig." At some point I this weekend I got this gash on my forehead, and people keep asking me, "What did you do to your head?"

"I don't remember. I just saw it the next morning," I reply, in all sincerity.

"Must have been a fun weekend!"

You people have no idea.

Funny, nobody has asked about the nice symmetrical rope marks on my arms yet.

I am really mad. I don't think I've ever been so pissed off at a business in my life. You have been warned.

I've been "working with" some people from our ISP at work to try and fix the abysmal availability of our internet connection. Since about thursday I've not seen it perform much better than 60kbit/sec. It's a wireless connection, 802.11b using a panel antenna thing pointed at a tower that's *maybe* a half-mile from us. The first dead chicken that their field circus waves, every time there is a problem, is to replace the antenna. So we did that. Didn't fix it.

After about a half hour of fooling with it, I determined that they have some sort of REALLY ham-fisted routing problem. It started with the discovery that one ping to anywhere beyond the gateway, including the gateway itself, gets me a random number of replies in the range of zero to five. DNS lookups using dig give bizarre errors about receiving things from unexpected hosts. The traceroutes to anywhere are just totally screwed. Traceroute probes each hop three times by default, and those of you who have even the most rudimentary understanding of IP networking will know that you should usually, if not always, get the same route every time, all other things being equal. Not so for us. In fact, not only am I getting a different router for each probe, I'm regularly, but randomly, getting responses from the same router at different hop numbers to the same host! Furthermore, tracing to the same host twice in a row gets a completely different output. Typically it's the same hosts, but in a totally new jumbled order.

I knew these people were inept from past experience, but when we signed up it was the only thing available. I should have known better.

Their tech calls back again on friday and says that he *thinks* it's because of congestion on the tower. Sorry, but if you're an ISP, shouldn't you KNOW whether or not your infrastructure is congested? That doesn't seem like something one would need to guess about. They say they're going to come over monday and move us to a differnent tower. I didn't get to take the call so I didn't get to ask him, nor did I get to tell him about the bizarre routing problem. Called him back, he's gone for the day. Wonderful.

I discovered on saturday evening that someone else is having the same problem with them, incidentally it is my friend's house, the network in which I put together and take care of. Same symptoms.

Later that night, back at work, I'm poking around trying to get some more clues as to what might be causing this problem, when I surf across a website that happens to display the source IP, which should be my IP. Except, it isn't my IP. I don't even recognize it, except for the prefix which I recognize as belonging to my ISP. I run down a proxy checker page, and sure enough, we're being transparently proxied! The fucking gall of these people.

I did some poking around and after 15 minutes I knew that they had installed some type of appliance that was doing the proxying. I could tell you who made the appliance, what OS it ran, what proxy software it was using, and what mail transport it was using--the thing is WIDE OPEN, on the order of 25 or so open ports with services running on them. It was even still using the factory configured hostname and self-signed SSL certificate.

It's bad enough that their service is abysmal. I tolerated it though, hoping that they'd get it worked out. I even tolerated their inability to get it worked out for awhile. Now I'm just angry, because in my gut I know what has happened. They've installed this appliance into production, without carefully understanding what it does, and it has hosed their routing. Now they are just hand waving the problems away, saying that our tower is congested and that they're going to put us on another tower. I'm angry because they've decided that we want proxying, and we have no choice in the matter. I'm angry because our activites are being recorded and cached in their proxy, representing a significant privacy problem. I'm angry because their ineptitude and negligence for security practice may put us at risk because if this appliance is breached, an attacker can have free reign to information about our activites, and furthermore redirect our connections anywhere he pleases.

I'm going to make the recommendation to my boss monday that we switch ISPs as soon as possible. It was one thing when their ineptitude simply inconvienenced us. It is entirely another that it is now putting us at risk.

Being sick blows. I am sick, have been since last Friday. It's this horrible sinus/chest coughing snot frenzy epizootie that everyone seems to have. My mother has it especially bad. We were both doing the zombie walk around the house most of Sunday and Monday, coughing continuously and almost uncontrollably. Dentist's office sent her home and rescheduled her appointment because, well, not only was she sick, but apparently so were half the hygenists.

Even the JoeyPan has it.

I coughed my brains out, tossing and turning, eyes ablaze and mucus hydrants WFO all of Sunday night and into Monday morning. I finally fell asleep, or more accurately passed the @#$% out, about 11am Monday. The strangest thing is that it's got this on-again, off-again thing happening. I woke up Monday evening feeling fine, except for a little drainage. My sleep schedule totally hosed, I stayed up until Tuesday morning when my head and chest went from semi-normal to DEFCON-1 in about 15 minutes. What the hell?

Even more hosed. Who needs a sleep schedule any damn way?

This is going on the 20th hour of my all-nighter to straighten out my sleep schedule, and having had a small nap, I'm doing quite well. I've managed to keep myself busy and piss away the day. Gentoo linux is pretty damn cool, if anyone cares. And a hot, hazy, shitty muggy day it has been. God I love summer. Air molecules, Erisian style. YAY!

Tonight the bed is really going to be calling to me. Hopefully that's a good thing.

Oh and it's Mexican Monday! Tequilla. (Yes, I know it's Wednesday. But Mexican Wednesday just doesn't alliterate.)

I got into a lengthy discussion about currencies and inflation with a friend today. I was trying to debunk the myth that inflation is nessecary to grow our economy, which I believe is not true. I thought I'd clean up my arguments and post them here, just for the hell of it.

The debate began with the assertion that there isn't enough gold and silver to cover what has happened to our economy. The logic was that if we eventually collected all the gold that was in existence, the economy would cease to grow.

"If there are 100 units of gold," my friend said, "and there are 50 people in the economy, that means 2 units per person on average. But if 50 more people move into the economy, now you are down to 1 unit per person. If a unit of milk costs 1 unit of gold to make, there are less people who can buy it."

"When more people move in," I counter, "the ratio of units gold to person becomes lower, effectively making gold twice as scarce. If Adam Smith is to be believed, supply and demand would dictate that gold becomes (for the sake of simplicity) roughly twice as valuable. Therefore gold will buy roughly twice as much 'stuff', it's buying power has doubled. Increased buying power means less of it is needed in exchange for a commodity. This is seen by our make-believe citizens as a drop in prices."

After considering this for a bit, I can't say I'm totally sure what the effect of this would be, but I think that the sheer number of people participating in our economy would be sufficient to smooth out any repercussions of this apparent "deflation" effect. The quantity of participants in an economy just doesn't make leaps of that magnitude in such a short time.

"But prices haven't dropped, they have always gone up," he points out, "it can't work in today's world." I didn't have the presence of mind at the time to point out many industries which regularly see falling prices. Electronics and computing are the most prominent example.

"Yes," I say, "but there is doubtful anyone in the US alive today who has lived in an economy with commodity-backed currency, and even if they did, they would be approaching 100 years of age, and their memory of it would be very sketchy I'm sure."

"Correct. We have grown out of the commodity-backed currency system."

"I would instead speculate that the prices of goods and services have been going up the past 100 years *because* of our debased currency."

"There has always been inflation, and always will be."

Again, it didn't occur to me at the time, but saying that an economy requires inflation to increase the amount of money in circulation because there isn't enough money to go around since prices continue to rise would seem to be a circular argument once one concedes that inflation *causes* prices to rise (I don't know anyone who would refute that). In fact, carrying this to its logical conclusion, inflation would spiral out of control, like trying to stay on top of a rolling log by running faster.

"I am not sure I agree with you there. But just a minute. Are you certain that prices of goods and services were going up in the late 1800s, early 1900s, when the US was on the gold standard?"

"Take a gun for instance. Are you willing to assert that the price of a gun was the same in 1700 as in 1900?"

"No, of course not. There are other factors involved in that example," I reply.

The intrinsic value of a gun fluctuated in that time. In 1700, all firearms were built one by one, by hand, by experienced gunsmiths. The laborious process and experience required to produce one meant that not many people could build them, and they were built rather slowly. This made them scarce, affecting the intrinsic value. Also, in that 200 years, the *use* value of a gun fluctuated, their use in warfare changed dramatically, the utility of having one in many jobs such as ranching and farming was discovered. All of these factor into the intrinsic value of a gun. Time for a new example.

"Ok," he said, reading my mind, "what about the price of lumber, coal, or anything else? The only thing that made prices go down over time was mass production of goods, in my opinion."

"Agreed. But consider this: Before the federal reserve act, a dollar was fixed at 1 troy ounce of fine silver. This was in the early 1900s. At that time, I have heard, a gallon of gasoline was roughly 25 cents. That works out to about 4 gallons of gasoline per troy ounce of silver. Today silver is a little over $10 a troy ounce, and gasoline is averaging a little over $2.50 per gallon, again working out to about 4 gallons of gasoline per troy ounce of silver."

"Thus no net difference," he observes.

"Between the value of the commodities, correct."

I went on to check my theory. I found at the US DOE a list of historical national average gas prices back to 1990. The price then? $1.19 per gallon. I then found a historical record of London Fix spot prices for silver. 1990's spot price for silver was about $4.85. That works out to 4.08 gallons of gas per troy ounce of silver. Again looking at my authoritative sources, July 10, 2006 national average gas prices were $2.91/gal. Silver? New York spot prices have been hovering around $11.40/oz in the last week. Doing the math, I found that I was indeed vindicated: 3.93 gallons of gas per troy ounce. Assuming (safely, I believe) that silver is a stable reference for value, the value of gas has risen a paltry 4% in the last 15 years, whereas the *price* has risen nearly 150%. The rise in value, if I had to take a wild guess, is probably mostly due to a cross between real supply issues, plus artificial supply manipulation, which jacks up the scarcity factor, counteracted by enhancements in refinement technology, which lowers the processing costs and increases throughput. But then again, it's probably more to do with the natural ebb and flow of the dynamic market that is petrochemical industry.

"So if the currency is fixed to a commodity," I continue, "the value of the currency remains stabler, and people don't lose the value of their money because of inflation."

"But that's not proven by your point. We aren't fixed to silver and gold now, so the system works. You are making a counter point."

"An inflated currency works fine as long as it doesn't *continue* to inflate, and this is why: Suppose you and I lived in 1900, under the gold standard. I have 100 ounces of silver in my pocket, and you have $100, which at the time, had equivalent buying power (remember, a dollar was fixed at 1 troy ounce of silver).

"Now let's jump in our time machine to today. I have my 100 ounces of silver, and you have your $100. But under today's prices, roughly $10 an ounce, my 100 ounces of silver buys 100 ounces of silver worth of goods and services. It is, after all, 100 ounces of silver. However, your $100 only buys 10 ounces of silver worth of goods and services. What changed? Well the *price* of goods and services went up, but as I demonstrated before, the *value* of them didn't. The price rose because the buying power of the currency diminished, thus it takes more currency to be worth the same value. My silver kept it's buying power, but your inflated currency did not." I forgot to mention that for the sake of argument, let us assume that the 1900s dollars were federal reserve notes and not gold and silver certificates.

Our conversation ended when he took a phone call from a friend, but I did continue to mull over the subject for some time. Another friend of mine pointed out something else: inflation is good when you owe money, because you owe the same amount of currency, but as it inflates, you owe less and less real value. Does 8.4 trillion dollars sound familiar?

I won't say that nobody can convince me that inflation isn't nessecary, but I haven't seen any logical evidence of this. As Penn Gillete said, a true skeptic demands to be convinced. So, if you think I'm wrong, convince me!

it's june already? damn .. luckily the weather is wonderfully hot. i can wear shorts and undershirts and get hot and sweaty. i'm at half-mast, really!

well except for crawling around in airplanes in the beating sun. that i could do without. meh! read this, you rascally gay republicrats (my emphasis):

Michael Badnarik, the only Presidential candidate to attend the San Francisco Pride Parade and Celebration in 2004 (the Libertarian, of course), has released the following press release concerning the current debate about same-sex marriage as part of his high-profile run for Congress this year. If only we could get such a clear statement from Hillary Clinton . . . but if I were you I wouldn't hold my breath waiting for one. Read at Outright Libertarians Blog

i hope he makes it to congress. that will be a big step towards forcing the mass media to acknowledge the third largest politicial party in the country.

random musings: firefox, python, xslt kick so much ass it's not even funny. oh and i am happier than ever that i have finally ditched windows for my daily do-stuff OS: 20 Things You Won't Like About Windows Vista

pesky politicians got you down ? .. check out the Politician Removal Service!

The new version of The Producers: "I'm in pain! I'm wet! I'm still hysterical!!" .. good stuff.

The Ice Harvest: Oliver Platt is the best shit-faced drunk i've seen in a long time.

i promised a visual. so here.



don't i look tough ? -- btw that's a rhetorical question.

I have a mohawk now. It's .. significantly longer than my usual annual summer mohawk.

So now i need some advice on what sort of axle grease/monkey snot/yak jizz i need to make the thing stand up nice and obnoxious-like.

Edit: pics are coming, you fiend (you know who you are).

Well. The end of April turned out to be quite the goat rope.. Internet connection at work went tits up for four days (i mean come on) .. floorpro showed up to start epoxying the hangars.. quite a bit of musical airplanes (not to mention musical junk-we-should-have-shitcanned-years-ago) involved with that .. began another sizable systems and networks project (sizable considering i am the systems and networks department) .. construction ramped up on my friend's house additions, i've spent probably every weekend in the past month over there, helping them fix trusses, modify floor joists, pull wire, and lay floor decking .. and, well .. i think that about covers it.

The good news is that, having worked myself into worthlessness, last week i managed to maintain some control over my sleep, woke up more or less when i intended every day. This week is not being quite so pleasant. I keep passing out after i turn off the alarm (i will never understand my brain) ..

Bad news? My to-do list hasn't shortened much. No gear time in awhile, truck still running .. "a bit peaked," as an unknown RAF pilot so eloquently put it .. I keep meaning to head up to louisville and spend some time with a friend but what with the weekend construction activity i've been rather unable. hopefully that will change soon.

$ cat >rant <<EOF
Ah yes, IML. I have the usual sprinkling of internet acquaintances asking, nay harping on, whether i will be attending. "It won't cost you anything!" .. I really can't explain why it WILL cost me, probably quite more than i can afford. Yes, even if i drive. Yes, even if i stay in your room. Let's just call it a gut feeling. So short of becoming blessed with some massive unforseen financial liquidity, no, i am not going.
EOF

Spring is nice, yes, but lets get with it here. Enough of this chilly nighttime BS!

I've always thought Eric S Raymond was a pretty smart guy .. But i found his blog today, and this entry really made me grin. Zing!

'Nuff said.

On an entirely different note... It is officially spring in the pissboy calendar: i have been bitten by some malevolent assraping arthropod for the first time this year. *scratch scratch*

I installed Xgl/Compiz on my dual xeon machine this week .. it is freakin' sweet, peter griffin style. Lick doo windows vista ! Quoth cbrus, "OMG you are a such a geek."



Make it bigger damn it ! .. Yes, it even works with video ..