Tuesday, September 28, 2010


It was inevitable, given that the whole "things happen in threes" doctrine seems, more often than not, to be an actual metaphysical rule of the universe. Or not, I suppose. I'm still waiting for the third. There must be some kind of expiration date on that kind of thing or it would be surpassingly easy to always match events into a set of three. We'll see how it goes.

So, what am I rambling on about? Well, it was one of those days when things that looked like they should be easy ended up not being so. It started at the paying job. I was putting the finishing touches on a new enhancement to a piece of software that we use to track a specific customer service process. The process in question has always been reactionary in nature, by which I mean we waited until a customer complained about a specific instance before taking action. These types of complaints are easily predicted, though, through a relatively simple analysis of data that we already have on hand. The reactionary approach worked fine for awhile, but the volume of cases has increased exponentially over time and is now threatening to overwhelm the lone person we have that works these cases.

Enter the IT dude. "Hey," said I, "let's shift our focus to a proactive approach. As part of my normal Wednesday morning data aggregation, I'll start tagging these potential cases and we can get busy on researching these issues before they get reported to us." Well, I didn't really vocalize that - I pretty much just jump in and do it. It turned out to be, as I had expected, a pretty straight forward process. There was the data, all lined up in a nice columned list on the screen. I put the data in the type of list that can be sorted just by clicking on the column header that the user would want to sort on. It was working great, mostly because it was re-used code that did the sorting.

It was tricky code early on - it struggled with correctly sorting values like $1.23 or 12/25/2010 because the computer (which I liken to a poorly trained monkey, with me playing the role of trainer) thought those values where words, not numbers. The most recent fix that I had applied fixed the "sort by date" problem by testing each value before sorting it to see if the DateTime object in my programming language could successfully convert the value to a date. If it could successfully convert to a date, well, I'd sort the values as dates. If not, they were numbers. Brilliant!

The new problem was with percentage values, more specifically percentage values of less than 10.0000%. Everything would sort just fine, except I'd see things like 2.2343%, 8.3423%, 4.3214%.  Clearly not correct.

Two hours of debugging later, I was mortified to find the offending line of code sitting directly under a comment that I had left in the sorting routine last time I had fixed it:

// I just want it known that I AM A GENIUS!!

I put that comment in there after (supposedly) fixing the date problem.  Who knew that the DateTime test was looser than Paris Hilton's overly hormonal chihuahua?

Thank goodness I'm the only one that ever sees this stuff!

All in all, it was a small-ish frustration tagged onto the end of what had been a fairly productive day. I thought I'd top the day off with a quick hour of work in the hangar. I scored a replacement for the lost headset jack nut for the RV-6 from the local mechanic and wanted to install that before I lost it, and I wanted to knock out the last page of Section 23. My hope is that the successful completion of Section 23, clearly one of the more demanding of the entire build, will help out when I get those 3:00 am anxiety attacks. You know, the ones where I wake up in a sweat wondering what ever made me think I could actually build an entire airplane! With my new response of "Hey, I got through Section 23, didn't I?" in hand, I can hopefully convince my middle-of-the-night critic to shut up and let me sleep.

The last page of Section 23 is where the first of the cowl hinges get installed. These cowl hinges have been the bane of my five years with the RV-6. I hate, hate, HATE these things. If you care to, you can find out why over at the Papa Golf Chronicles by clicking here. Or you can just take my word for it: the manner in which the typical Van's RV attaches its cowls is an unmitigated pain in the, uh, rear end. The RV-12 is no different.

It seemed a simple job. Cut off a couple of 10" lengths of the hinge material. Clamp a couple of the pieces onto the forward flange of the lower firewall. Match drill #30. Rivet in with the ubiquitous LP4-3 rivets.

I should know by now, and I suppose I do at some repressed subliminal level, that jobs that look easy most often aren't. It was a rather odd set of problems that I ran into. The first problem arose when I tried to clamp the right side hinge piece to the firewall flange. The firewall seems to be under some sort of compression force that makes it "oil can" (for want of a better term) in and out. In other words, the flange would either be pushed into the fuselage, or popped out of it. It wanted to be anyplace but aligned with the holes in the fuselage side skins. I couldn't cleco it or clamp it in place because the hinge strip wouldn't be able to sit flat against the flange with the cleco stems or the clamps in the way. I tried all kinds of things to keep that firewall from popping in and out; none worked. I eventually resigned myself to the fact that I'd have to start at the top of the hinge strip and clamp it down over the cleco stems in the middle set of holes. Doing that put a bend in the hinge strip, but I decided that the bend would be okay once I get the hinge fully riveted in place.

That turned out to be correct. Unfortunately, I was so focused on keeping the firewall flange in alignment that I let the hinge strip drift away from its intended position of being flush with the front edge of the flange. Down towards the bottom, the hinge strip angled in towards the firewall. That would never do; I know enough about these cowl hinges to know that they have to be pretty well aligned. I'd have to start over with a new piece of hinge strip. But I'd defer that until I had the left strip clecoed in. That's when I discovered (and I still can't figure out how this is even possible) that I had cut four right side pieces and no left side pieces. 

Normally this is when I would have started worrying about not having enough of the hinge material to correct those two mistakes, but I have plenty of the stuff left over from the repair to the RV-6 cowl. I cut a left side piece from the remaining supply. I had hoped that the left side of the firewall wouldn't do the same oil can thing that had plagued me on the right side, but to no avail. Still, having figured out how to address the problem helped get the left side done fairly easily. Going back to the right side, it too went as well as could be expected. All of this effort ended up taking more than two hours.  So much for an easy night at the shop!

But... I slept through the night without so much as a pang of anxiety.  Goodbye, Section 23!

1 comment:

Marc said...

I had to chuckle because I've also been known to write self-promoting comments. Must be something we programmers do. :)

Post a Comment