The fan is in place. Here comes the shit.

[roothorick]

What a delightful day today. Marketing Boss wanted to do some graphics today. That didn't happen because...

• What was supposed to be a 30-minute POS terminal replacement turned into a 7-hour ordeal that involved: adding pieces to the new terminal that were supposed to be shipped with it, but instead had to be salvaged off the old terminal, pulling the motherboard out of the new terminal and changing jumper settings on it (by the way, these jumpers were in a really awkward position, it took all the dexterity I had to set them right with a needlenose), and a myriad of oddball software issues caused by it being set up as Terminal 1 when it's actually Terminal 3. It wouldn't have taken so long if I didn't have to constantly play phone tag with Impale support.
• Laser tag has developed a problem that, I'm not kidding, has the Lasermovies main programmer stumped. And it's such an issue that if you push the wrong button, a whole bunch of critical stuff stops working until you do some non-trivial maintenance.
• Flip-2-Break, #2 in our gameroom, simultaneously developed three separate problems that put all of both doors completely out of commission. I hope we don't actually lose revenue over this.
• One of our skeeballs blew a sensor. It just goes straight to 900 when you coin it. I turned it off and ignored it. I'm not dealing with a machine that takes a small team to move when I've got a major attraction with major problems.

All on a friday afternoon, when things are getting busy, parts departments close early, and everything is under constant attack for three days straight. I'm NOT going to work this weekend.

[roothorick]

I should probably elaborate on the title a lil bit.

Laser tag broke in such a way, if you push a certain key you're supposed to push during a normal game cycle, the little glue program that connects the game computer to the reservation software explodes, and the only way to get it working again is to dig into an obscure directory and delete some files. We put a sign up to not push that key, but how much you wanna bet, at 10:30 after we open and do our first game, I'm gonna get an angry phonecall from the laser tag desk?

[roothorick]

Laser tag STILL isn't fixed. The Lasermovies programmer, let's call him Columbo, has never seen this issue before, and that's quite obvious. The process has devolved into:

1. I get an email from Columbo telling me to do X, try Y, try it again, and send him a bunch of files from the reservation computer running the printer driver program.
2. I go over to laser tag, do what he says. Try to print, doesn't work.
3. I send him an email saying no, it doesn't work, and the files he wanted.
4. HOURS later, I'll get another email, and have to go back to work. The cycle starts anew.

Tomorrow will be Day 3 of We Can't Print Scoresheets Period. And there's no end in sight. Big Boss is NOT happy.

[teh_blumchenkinder]

Ugh, I'm getting a headache just reading this!
I feel for ya, man! (Not that that will help much... )

[Rapscallion]

Mended yet?

Rapscallion

[roothorick]

On monday, I fixed it completely by accident. By breaking it.

I'm copying this straight out of the e-mail I sent to Columbo. A little explanation: Columbo sent me debugging versions of the glue program to try to sort the problem out, called test, test2, and test3. These debugging versions pop up dialogs as they go through the process of processing and printing scoresheets. The second set of dialogs, as far as I can tell, was the program starting to process a job again while it was halfway through processing that same job (the first set). Somehow breaking it this way magically fixed whatever it was getting hitched on before.

I started test3, and clicked "check jobs".. a series of dialogs started, then I got a second set on top of it.. I clicked through the top set, and it actually printed! The first set failed with a different error (file not found)... but now it's working, no more crashes. A race condition, I'm guessing, triggered by some state data stored on disk somewhere. And somehow, processing the same job twice simultaneously in separate threads corrected that data.

But I don't have the source code, you do. You'd know better than me.

In hindsight, I do NOT want to see that source code.