Category: Computing

In the Wild

It’s a little wierd now that I’ve been working on actual published software for a few years, as opposed to research, to see actual products I’ve worked on in the wild, and to read reviewers’ comments where I know exactly what the software in question does, because I wrote it.

Andrea & I have just put our place on the market, and are looking for a new place, and we’ve been getting lots of real estate listings from our agent.

What makes this relevant is that a couple of years ago, during a lacuna in employment between computational linguistics research and game development, I worked for a company whose software is used by half the realtors in North America. And indeed it turns out that the printouts we’ve been getting were made by that software.

Now I’d like to say that I know exactly what the software that made those printouts does, but the software in question is a vast and terrible jungle of recursive and self-modifying spaghetti code without end. I must confess I quit that job partly out of fear of ever being asked to work with it.

I much prefer physics and artificial intelligence.

Disappointment

Dear ACES Team,

So I’ve finally got FSX running a somewhat decent framerate, thanks to all the helpful hints on the net. The biggest improvement was due to turning antialiasing off and filtering down to bilinear and letting the video card do the rest.

So now I’ve got consistent frame rates of ~25 fps, even with normal autogen and 25% airline traffic. So I embark on a flight in the 738 from CYVR to CYYZ.

Couldn’t someone at ACES take the time out of their elephant-producing schedule to adjust the one variable to make ATC hand you off at a rate somewhat less than once every 30 seconds? And the other variable that make them clear you for a higher altitude 300ft before you reach your assigned?

But ignoring these annoyances that have survived two major version updates (in favour of Elephants!), the flight goes more or less as usual. There’s the awkward fact that every 5 seconds the entire terrain and clouds blink in and out of existence, but I guess I can live with that, because Bill Gates called me personally to assure me it’s all in my head.

Until I’m trying to intercept the ILS for CYYZ 33R. The GPS and the map both agree that the freq is 110.30, so that’s what I’ve tuned. But no ILS ever shows up. So I take over, do a bit of a sideslip over to the glide path (by the way, what freaking genius thought that the maximum bank angle for the 738 on autopilot should be 10 degrees, with no way to change it?).

At decision height, frame rates suddenly drop from 24 to 0.5. I kid you not. No gradual slowdown. All of a sudden I’m watching a literal slide show. I did manage to get on the ground:

So far my FSX experience has been one unmitigated disappointment. Not just this flight, every single one so far has been a exercise in complete frustration. I’ve been using the product since version 1.0 in 1985, and this one has given me the most grief of all of them.

To those who say “don’t expect to get all the good new stuff with old hardware”, I say bull pucky. At the same level of settings as FS2004 (the actual numbers in the cfg file, not just the sliders), performance is far worse. In any supposed “new version” of a product, I expect at least the same level of performace as the old one, given the same requirements.

The autopilot is even worse (I know that’s not saying much) — altitude and course excursions all over the place, and the aforementioned 10 degree bank angle. There are absolutely zero improvements to ATC — couldn’t they take even a day out of their elephant-producing schedule to tweak the two variables controlling handoff times and altitude clearances?

So thanks for nothing, ACES. You may have made some kick-ass elephants, but as far as my perception of your dedication and abilities to make an actual product that works like it should, all I can see is a big pile of elephant doo-doo.