In case you weren’t following our tweets yesterday, we both completed the ride in reasonably good time. Nat’s motionbased report is here, and mine is here.
A few things:
- This was only the second or third ride with my new seat. Normally this would be a bad idea, but there was no way I could have done the ride with the old seat. I love the look of the Selle, but my ass (along with my naughty bits and, strangely enough, my feet) love the Toupe.
- This was a check-up ride for my RAMROD preparation. I basically wanted to see if I had any hope of riding the mileage and climbing (154 miles, 10,000 feet climbing) coming up in 10 days. I think given how close the numbers are, I’ll be okay. Also, this ride was in some ways harder than RAMROD, I think. All my training in SF has been on long, steep, sustained climbs (which seems to match fairly well with the RAMROD profile), and the climb to the clouds course was just roller after roller after roller. Difficult to pace yourself when your speed is constantly yo-yoing, and your heart rate is as well.
- What the motionbased info doesn’t show is the fact that I rode through a torrential downpour about 10 miles from Cambridge. This is probably why my elevation dropped to below sea-level, my cadence dropped to zero (I *was* pedaling), among other little problems (the gps unit was constantly flipping between auto-pause and auto-resume, while I was going 21mph). The downpour left me completely soaked, and while the headwind was at times a little chilly, the cold was much prefered to the heat of the day, and it gave me my 4th or 5th wind - enough energy to close out the day strong instead of limping across the finish line. It also helped to wash a lot of the salt out of my shorts, so I wasn’t too embarassed walking into the hotel.
- Cookies at rest stops are amazing. Screw the fruit, bring on the oreos. Also, I hate gatorade.
Been quiet here for a while. Been either working or training hard for the rides that are looming in the next couple weeks.
I’m off to Boston in a couple hours to put in some long miles on flatter terrain (oh, and work a little :), and ride in the Climb to the Clouds century ride (~145ish miles from my starting point in Cambridge) put on by the Charles River Wheelmen. I rode out to Mount Wachusset a couple years ago in October, froze my ass off, and even ended up on Route 2 (not 2a) for a little. It was a miserable experience, so following a cue sheet seems like a sound plan for my revenge 
After hopefully around 300 miles on the bike during the week I’m in cambridge, I’ll be back in sf for 5 days of relaxing before driving up the coast to Seattle to ride in this year’s RAMROD event. 154 miles, 10,000 feet of climbing. It’s gonna hurt, but almost everything worth doing hurts a little, doesn’t it
More sporadic updates here to come, and more up-to-date ones on twitter (when it’s up).
I’ve been putting in a lot of miles lately. Mount Tam (at ~50 miles and 5000 feet of climbing) has replaced Conzulman as my almost daily ride, and I have a nice little google calendar set up with mileage/climbing goals to get me to within 80% of the RAMROD mileage/climbing by the time the ride happens. Not a lot of wiggle room, though.
A few things, in no particular order:
- I thought I’d seen the height of badassery when I passed a man with tree trunk thighs, leaning way over his handlebars climbing up Conzulman, on a fixie. But today I saw what might just be more hardcore: A guy climbing that annoyingly long hill up from Sausalito to the bridge, on a unicycle.
- Saturday when descending from the east summit of Mt. Tam, I came very, very close to running over the head of a very, very long snake that was sunning itself on the road. I jerked my wheel to the left (at 30mph) and narrowly avoided it. I didn’t actually see the wheel miss it - it all happened so fast - but the front tire didn’t show a wet or bloody spot, so I guess it survived.
- Most people remember the road rage incident in Australia last month, but Peter showed me something quite a bit more horrifying today, thankfully after I returned from my ride, or I might have never left the house. Pretty intense (and wholly amazing from a “I can’t believe he captured that image” perspective) carnage
here. (Update: it appears the flickr page has been taken down, so against all reasonable copyright law, I saved my cached-copy and uploaded it here: cycling-accident.jpg)
I’m so glad I don’t act on half the ideas I have. If I did, not only would I get much less done on any one of them, but I also wouldn’t be pleasantly surprised often that someone had the same idea and got it to a state where I don’t have to act on it at all.
Today peter pointed me at SugarTracker
Over on Steven Engelhardt’s blog there’s a mandelbrot viewer written using silverlight 1.0 (meaning just xaml + javascript). In his post he listed a few problems he had, and one caught my eye:
- I can’t figure out how to do progressive (e.g. line-by-line) rendering to give the user feedback.
I figured this would be pretty easy with the help of setTimeout() and some more global state, so I hacked it together here. Instead of the line by line rendering I implemented a progressive jpg style interlaced rendering, with multiple passes of smaller and smaller pixels. As you can see in the demo, you can mouse over and click while it’s in the middle of rendering, and it’ll start the new render. No more waiting.
One thing he’s doing (that he notes is quite heavyweight) is using rectangles. To lessen the hit, he uses 1 rectangle for many contiguous pixels of the same color on the same line. I’d imagine using an InkPresenter and adding stylus points might be a less heavyweight approach (at least I think it will be in the case of moonlight.)
update: realized how to make mouse events faster - specify IsHitTestVisible=”false” on all the rectangles. That way the event system doesn’t have to worry about generating enter/leave events for each scan line (and for many individual pixels on those scanlines). And speaking of scan lines, add an translucent red indicator line that shows the next line to be drawn. Add more precision (down to pixel_size == 1), and when we’re finished rendering, display how many rectangle children of the main canvas there are (for the first render there are over 13000. that’s a lot of children. moonlight seems to handle it ok too
