Wednesday, October 18, 2006

Exceptions

Wednesday: Woke around 8:40. Down basement for brief sitting and hello to friends. Cats. Got back in bed, but didn't sleep. MB headed out to an appointment with a banker. Up, shower, and so on. Email. MB returned, working at home today. I put the hammock and tiki torches away. Printed some flyers, and dropped them back at the house. Put some CDs in the mail to a friend. Off to work.

Working on the add-on issue that had been reported by a developer migrating away from our API, I found there was indeed a problem on our end. Nice fellow. I spoke to him in an overseas phone call.

Took a walk out and got some cash, supplements, an rx, pizza and coffee. The coffee has gone up in price by 15 cents, but they're giving us promotional mints. I offered back the mints if they would give me 15 cents.

Turned my attention back to the difficult to debug problem I'd been looking at last night (a by hand binary search of possible throw locations, since "break on all C++ exceptions" doesn't work in XCode.) Mrk suggested simply setting a low-level breakpoint at the library routine. After some head-scratching, we managed this, and within minutes I had found my bug. My question is, why doesn't XCode just do what we did, as its implementation of "break on all C++ exceptions?" Very weird. Fixed a satisfying and long standing bug with a very important provider-dependent class, with this new information. Started looking at some sort of possible better behavior for the add-on issue. Came up with one possible solution, but it didn't do the whole job. About that time the USB on my computer started flaking out (because I keep hot-swapping between Intel and PPC). Headed home around 10. Still no word from R&A, who were supposed to have landed this afternoon.

Back home, MB had just returned from a shopping trip. Helped her put away groceries. Some soup and potatoes. Email. Now some guitar and completing the day.

No comments: