CodeRace 2004: Day Two

[maradydd]

Headache gone. Beautiful day outside. Haven't left the house today.

Authentication works perfectly now. (Had some grind-your-teeth issues with a Heisenbug which took an hour to find and left me feeling intensely stupid.) As soon as Alex gets out of the shower, I'm breaking for lunch, after which I will write scripts for:

- give-me-a-random-abstract EDIT: Done, though I had to reload the DB to do it - who knew you couldn't add serial columns to an existing table?
- keyword search EDIT: Done. OMFG, tsearch is fast.
- evaluation EDIT: Pretty much done. Can't be arsed to recreate that one really complicated join right now, but as soon as I do it, deletion will be as near to finished as makes no odds as well. I sleep now.
- delete-previous-evaluation

At which point I will still be right on schedule, and will either call it a day or keep going.

Reason why it's good to have low blood pressure: after just losing two months of work to a hard drive crash, suddenly hearing one's laptop's hard drive chattering unexpectedly. (And then finding out it was just a cron job. But still.)

More news as it happens.

EDIT: Total lines of code written today: 252. Would have been more like 300 if I had buckled down and worked after dinner, but I didn't, and that's my own fault.

Comments

[oralelk.livejournal.com]

Go you!

[steer.livejournal.com]

heisenbug?

I'm guessing one of those memory leak type errors which are particularly hard to track down?

[maradydd.livejournal.com]

More generally, a heisenbug is any bug for which you can determine its effects, but not its location (akin to being able to determine a particle's location or velocity but not both). Typically, when you try to reproduce the code elsewhere, it works just fine. I suppose you could also have a heisenbug where you knew its location but not its effects, but that would be more like a heisen-undocumented-feature.

Memory leaks are definitely a kind of heisenbug. This one was actually a db error being thrown and caught correctly, but generated in a different try block than the one I thought was responsible for the error. Cue lots of tearing of hair, gnashing of teeth, &c. as I replicated the try-block I was looking at in a test script and it worked just fine...

[steer.livejournal.com]

(grin) That's pretty much what I guessed the definition must be... good word, I like it a lot.