It was raining inside and out at work tonight. I made eight code commits in the space of an hour. Such commit storms are pretty typical for me, but tonight's was eight commits for a reason. The last one had a revision number of 314. Sure, it was late for Pi Day, but I wasn't interested in staying around long enough to reach 501 for May Day.

I've been using PNM files for many years. The earliest use I can recall was the Digital Stump. More recently, I've treasured the PGM portion of the spec as a super-simple format for grayscale raster image data. The format uses a text header that can be generated with a single fprintf. Plenty of free software is happy to accept 8-bit PGM files ( Read more... )

( C++ exceptions mystify me: Cut for technical content )

( A fatal web usability flaw in my own project? Argh! )

( Cut for the folks who can't handle potty talk. )

I'm working through more online Six Sigma training at work. One of today's topics was the Five Whys. If something's wrong, ask "why?" and write down the answer. Think about the answer and ask "why?" again. Lather, rinse, repeat. By the time you've asked "why?" through five levels, the root cause is often apparent. I laughed when the course ( Read more... )

This afternoon, it became apparent that our project needed one of our old Sun Ultra 60s to be upgraded to Solaris 10. I promptly marched off to discuss the matter with our usual IT contact for such things, Vicki. Vicki asked if there was anything on the machine that needed to be saved. Knowing it was not a machine we usually used, I said no, but ( Read more... )

One of the things we learned in training this week is a FMEA: Failure Mode and Effects analysis. It's a template for capturing a brainstorming session's worth of failure scenarios. We were split into teams to go through the process. One of the template's columns is "severity". Some folks start their discussion with the highest-severity ( Read more... )

A big chunk of our training this week was about understanding statistics and measures. Sure, you may know the mean, but if you don't also know the shape of the overall distribution, do you really have the whole picture? Is the sample size good? You may recall that back in September, I was guessing 90% mastery of the material during that training ( Read more... )

Today I learned that a histogram is not equivalent to a bar chart. A histogram's main axis is always a bucketing of a continuous variable, whereas a bar chart is typically for counting discrete categories. Am I the last one to be disabused of this notion? We also talked about Pugh matrices. This is a nifty trick for assisting in selection ( Read more... )