It's good to have a rfrench

[cjsmith]

Rob fixed the furnace.

I like to think that my suggestions about isolating the individual problems and about the possibility the board itself could contain the flaky connection were useful. Either way, I'm very VERY glad we have working heat again. (For now.)

Comments

[psi-star-psi.livejournal.com]

I was unsurprised by Rob's entry when I read it. After watching him spend 4-5 days beating into submission the most evil bug I've ever heard of (tied with one that took Brad & Chris with others 3 days to crack), I don't believe in a problem that he cannot crush. It just requires sufficient motivation.

[cjsmith.livejournal.com]

Yup. And compared to most software bugs, this one was really quite simple. It was awkward to look at for mechanical reasons but it wasn't fundamentally a difficult problem at all.

[rfrench.livejournal.com]

Which bug was this?

[psi-star-psi.livejournal.com]

This was the Boolean size discrepancy bug from GDC bringup (tools and driver used different definitions). I know that it was probably not memorable to you, but I still have no idea how you managed to isolate it. The same with the one Brad and Chris cracked, where inconsistent implementations of 64-bit shift were linked in from different libraries. Gack.

[rfrench.livejournal.com]

Actually, now that you mention it, I do sort of remember that bug. I also remember the 64-bit shift one, although I didn't really help debug that one.

[cjsmith.livejournal.com]

Oh man, I remember that one. (The Boolean.) I also remember the Andrea-rewrote-booleans-as-a-class bug, where querying the value of something wasn't atomic. That gave us the fan effect.