Apparently the counter bits are still having problems, which just freaks me out since the bottom 11 bits don’t have any problems at all. Tracking the b0rkd bits, they seem right about 99.44% of the time, but are wrong often enough to be a semi-obvious bug. I don’t really wanna have to rip this crap thing apart again and checking impedance through the solder points, but I’m beginning to wonder where else a problem could be?
So the LUG meeting went well. A couple of new people didn’t quite seem to understand that staying quiet while we’re trying to listen to a presentation can be a good thing. Ugh.
I hate people that make broken hardware asking me to workaround it in software. I really really do.
Lunch at a place that supposedly has really good barbecue today. We’ll see.. it’s been like half a year since I’ve had any, so it’d be a nice change (I believe they have to kick me out of the state if it goes much longer).
I’m mostly installing server and business-systems machines, so they don’t have video cards of the “Super Rage Death Fury Evil Twister 3000 Plus” variety — Graham Mainwaring
After having to jog from bldg 002 all the way out to the Davis Drive entrance of IBM (ugh, longer than it looks when you drive it) to let the guy (now it’s officially more trouble than if I had just replaced the tire myself), I finally have 4 working tires and can drive my folks to the airport (although they conveniently forgot to let me know when they were leaving Fayetteville!)
The whole security-gate fiasco loses me the half hour that my parentals were waiting at my house (at least enjoying the nice day on my deck, but still).
Lunch at Penny’s was yummy as usual. Got the Orange Tootsie Pop, narrowly sneaking it away from Nitin (I saw him admiring it, I swear!). Sneaky bugger.
“30 to 60 minutes” for the Ford resp. time… we’ll see.
LinuxToday: Switching subjects, I understand you like to watch Scooby Doo
Alan: Rather a lot, yes
LinuxToday: Every episode?
Alan: The stuff with Scrappy Doo is best avoided, I think
So I’m driving in to work and coming up on that place where I-540 merges down from 2 lanes to 1 before swerving off to I-40. I have to speed a little (95, only 10 over the 85 everyone else was going) to get past this semi when the expedition starts shaking a little. The alignment’s less than perfect (Jessica needs to stay off those curbs when I let her borrow it 🙂 anyway, so I shrug it off.
Getting to a slower speed going through the gate at work, I notice it sounds like something may have gotten lodged in the wheel well. Looking around when I park it, I notice it: a huge (the tires on this thing are ridiculous) flat tire. Ugh.
Because it’s still under 36k miles (awfully soon for a flat), I’m obligated to call this silly 800 number Ford has and have to change it. I don’t feel like dealing with this, or them, today since I’m still catching up on some of my Swing+JDBC work, but there’s not a lot I can do about it now.
The only nice thing is they can do it while I’m out at lunch anyway (since there’s no need to actually get in the vehicle), so maybe it won’t be an entirely lost day.
Stock I’m happiest about getting out of at the right time:
Bleah… I get the feeling this guy is screwed since he hosed up the raid superblock for one of his partitions… guess I’ll see tomorrow, but since I’ve got a working test suite for the data streaming library, I’m Audi here.
It’s past midnight and I’m still here at work helping some random loser recover his Linux software raid… what’s wrong with me?
Well, it turns out adding other record types actually forced me to split the support code in two, actually making the data stream handling much cleaner. I’m going to cave in and pass the increment amount as an argument to advance_stream simply because for most sane architectures (in this case, x86 included) the first 4 args are going to be in registers anyway… saves a load from occurring, although one we’re guaranteed to hit in the L1 anyway… shrug
Shoving the record type definitions off into another file is also nice since I can let Dan (and other developers as needed) edit that as record definitions change down the road…. should make life a little easier ones we settle on a format for the records in the ia64 traces…