I read a bunch of zombie themed novels. Some have lots of zombies, others not so many. Some are real books, some are what I was hoping were the upper echelons of more or less self published work.

more...

At some point Time’s humor columnist, Joel Stein, transitioned to writing about more serious topics. Still funny (if you thought he was funny), but less fluffy. The March 18 issue (Sheryl Sandberg on the cover) is a good example. It’s not yet online that I can find, but there’s a not funny similar article, albeit with a different conclusion at the Guardian. Do online comments hurt – or aid – our understanding of science? Stein also refers to these numbers about the Guardian’s comment stats.

He makes a great point towards the end of the column about how adding comments affects Time’s reputation. Generally negative. I’ve noticed the same with several newspapers I read. Why do they have comment sections? As Joel says, about the only thing the comments discuss is “whether the President is a horrible communist or a terrific communist.” How does the newspaper gain from reserving a part of every page for idiots? Is the all important engagement metric aligned with what they want to optimize? I try not to read the comments, but sometimes scroll down into that region by accident, and then I’m stuck reading them. And then I generally close the tab because I realize I must be reading a newspaper written for morons, and somewhere out there is a better website, a website I should be reading. Are the people who comment really of higher value than the people the comments chase away?

Here’s the setup: Heartless prosecutor offers kid a deal. Take the plea bargain and go to prison for one year. Or go to court and face thirty years. Inspired by true events.

Features Omar as Omar and the Rock as not the Rock.

If you increment a signed 64-bit counter at 1GHz, it will take about 300 years to wrap. 600 years for unsigned.

OpenBSD 5.3 is coming soon, the order page is up now. I don’t think I made a single commit for this release, which represents a new low. At least things can only get better, I already made a few minor commits for 5.4. One feature that I did hack on but which may go unnoticed because it’s largely invisible is a new disk I/O sorting algorithm. beck@ was the driving force in testing this and pushing me to write the code he wanted. :)

more...

My HP Mini has an unsupported wireless chipset, so I was in need of a USB adapter. To match the host system, it too must be mini. I started with a Keebox W150NU I had sitting around. Works well with the run driver, but it sticks out a little, so I picked up the B-Link BL-LW05-5R and the Edimax EW-7811Un on the cheap. Both use the urtwn driver.

All of these operate solely in the 2.4GHz band. Local download speeds are generally 1.5-2 MB/s for the Keebox and 2-3 MB/s for the urtwn devices. ifconfig reports the signal strength in dB, but I’m not smart enough to know what that means. No issues with reception moving around between rooms. Occasionally, while using a urtwn device, I’ll see complaints from ehci about such and such not being just right, but it doesn’t appear to affect performance. The Edimax has a glowing blue LED in it, the B-Link has a rather obscured LED on the end. I don’t need another blinkenlight poking me in the eye, so the B-Link wins this round. Amazon tells me I paid $6.70, shipped.

From left to right: Keebox W150NU, B-Link BL-LW05-5R, Edimax EW-7811Un.

Yesterday, Windows Update decided I won the IE 10 lottery and earned myself a new browser. Then I remembered the IE SunSpider performance scandal (turns out this was for IE 9, my how time flies!) and my own thoughts on the difficulty and fragility of dead code optimizations. For background, the Lies, damned lies, and benchmarks article at Ars is a decent read.

more...

The only thing more fun than talking about yourself is talking about the blog platform you use to talk about yourself. I rolled out some flak upgrades today and it almost worked. I did learn one critical lesson about sqlite which may be interesting, after that the post devolves into useless trivia.

more...

I signed up for Amazon’s Mechanical Turk service a little while ago. As a worker. My theory was, I’m sitting around watching Glee, but there’s all this plot and drama stuff I don’t care about happening between the Journey song performances. I could read or code or something, but then I get sucked into that and miss the song. A micro tasklet I could complete in a few seconds sounded like just the thing, and making five cents a minute was five cents more than I normally make watching TV.

more...

As part of working on mailtanium, I wrote a basic webapp in Sinatra to check mail on my phone. Sinatra does all sorts of magic for you, as ruby frameworks are prone to do. It’s built on top of Rack, which also does magic things.

I wasn’t fully aware of this until I recently checked my logs and noticed a bunch of entries like this.
W, [2013-03-06T06:01:57.276947 #2149] WARN -- : attack prevented by Rack::Protection::HttpOrigin
Wowza, I’m under attack! No wait, on further inspection, I see one of those every time I send a POST request from my phone. Firefox on my laptop doesn’t seem to trigger it.

What’s happening? I’m not sure. From what I can make of the source for the module, the request should be blocked, but it’s not. Everything still works. The emails I tried to send were, in fact, sent. Maybe Rack’s default config is to only log a warning and not do anything drastic. But why then say the attack was prevented, instead of not prevented?

I think what I need is less magical protection and more protection from magic.