pleased precariously

does anyone know if Xvfb short circuits out input events (would make sense, I suppose, seeing as how it doesn’t even display a window)?

I seem to be unable to replay xnee recorded sessions against an Xvfb server and have it do the things it needs to do. The windows the session interacts with are in the same location as when they were recorded (using an Xnest server), so I’m at a loss to explain why they’re not working against Xvfb..

any ideas?

So, like others I was kinda put off by the sdtimes article. To think that our System.Windows.Forms implementation has reached the point where the code churn is manageable enough to not ever regress is just foolishness. We’re getting there, but it’s disingenuous (and a little self serving) to imply that it’s some lack of process that’s causing regressions.

Anyway, after reading the article, I did my part to destabilize the build for a few hours before heading home to slowly tie one on with perhaps the greatest rum ever distilled.

It might be the alcohol, it might be my arrogance, but I decided I could hack up an automated UI testing framework, using Xnest, Xnee, xwd, and diff.

So I did.

Steps to glory:

one time setup (per test)

1. Start up Xnest -geometry 1000×7000 -ac :1
2. run “mono swf-testboxes.exe” with the DISPLAY=:1
3. run “cnee -display :1 –record –mouse –keyboard -o textbox.xnr”
4. do some stuff, like c&p between textboxes. Ctrl-C the cnee.
5. xwd -display :1 -root -out expected-result.xwd

test run

1. run “test.sh swf-testboxes.exe < test-script”
2. note whether PASSED or FAILED is printed to the console.

test.sh is:

and test-script is an alternating series of lines, following the form:

xnee-playback-script
result xwd

In this case, it’s:

Where testbox.xnr is what we generated above in the setup step, and result.xwd is prepended with “expected-” to form the name of the file we must match.

It’s all pretty rough, and just because we have a means of automating this stuff doesn’t mean that we’ll suddenly have a bazillion tests.

If you’re interested in helping out it’s pretty easy. Grab the xnee source and get it built. Make sure you’ve got the RECORD extension enabled on your server. Start writing little mono executables that test particular features and fixed bugs. Record yourself using the executable (using the behavior fixed in the bug), and mail me the .cs, xnee playbackscript(s), and result xwd(s), and the bug # it’s meant to test. It’s that simple.

I’ll see about generating some this weekend.

If the San Francisco trip showed one thing, it’s that when I’m on my bike I’d rather be climbing. Flats are boring, descending is tons of fun, but give me a mountain to go up and I’m happiest.

So, I’d been looking for something like the ride Alex and I did a couple weeks ago (keeping in mind I don’t have a car out here). I saw mention of Mount Wachusett, and even though it’s a much longer ride (easily twice the distance as the ride Alex and I went on) with roughly the same overall elevation gain, I was intrigued. I shelved it for later thinking. It was definitely something I wanted to do, but I could wait a while…

I was supposed to do Boston to Provincetown yesterday. I’d borrowed Nat’s eTrex with waypoints and street maps so I wouldn’t get lost like Rony and I did last time we tried. But when I woke up at 4am to find the temperature hovering somewhere in the 40’s in my bedroom, I decided to stay under the comforter for a little while longer and hit the off button on the alarm. When I finally did get up at 6am, a quick calculation led me to conclude that I’d miss the 3pm ferry back to Boston. I didn’t want to leave boston at 7:30am and get back at 8:30pm - that’s just too insane for me - so ptown was out.

I broke out gmap-pedometer, found Mt. Wachusett, and set about figuring out how to get out there. A portion of the route I drew up went on Route 2, but the bike map I had showed a big bold route 2 branching off from there, and I figured the smaller route 2 would be okay for bikes.

Of course, this was insane. Nothing like semis blowing past you at 70mph to get your heart going. As soon as I saw the first Exit sign I knew I’d made a mistake, and got off. One thing about riding near traffic moving that fast, though, is that the draft is amazing. Pedalling along on a flat at 27mph is pretty insane anyway, but doing it at such a low heart rate is incredible. It almost made up for the broken glass and pieces of tire I had to avoid while on the shoulder.

So now I’m somewhere in western mass, near Acton, off the cue sheet, without a map, and without a clue where I am and how to get where I want to go. I learned two things at this point:

1. If you want directions as a cyclist, go into a gas station and start looking through their maps. If the guy behind the counter doesn’t beat them to it, a customer will ask “where you heading?” and then provide you with all manner of help. And suprisingly, most of it is very useful.
2. Don’t think in street names. Think in route numbers. Telling someone you’re looking for “Princeton Street” isn’t nearly as informative as telling them you’re looking for route 62. Also, a given route changes names at least 5 times over the course of 10 miles, and most road signs use the route number, chances are whoever you’re talking to knows the street as a different name anyway.

It turns out my complex cue sheet could have been replaced with about 4 lines:

• Take 117 to 110. Turn left on 110.
• Take 110 to 62. Turn right on 62.
• Take 62 into Princeton Center. Turn right on Mountain Road.
• climb mountain.

So anyway, after the Route 2 fiasco, I found that the guys at the 2 gas stations and 1 garage sale had steered me along the above route, and also happily found that my cue sheet followed the same route. It made me feel better knowing I was going roughly the way I wanted to, even if it was totally by accident.

By the time I got out to Princeton, I was hurting. The climbs were getting steeper, and I’d already put 50 miles in. I had it in my head that the climb would be like Mt Tam, where basically you went from flat to climbing, and the climbing didn’t really stop until the top. The route I took up to Wachusett started climbing about 18 miles away from the summit. Steep climb followed by a shorter decent, steep climb followed by another shorter descent, etc. This 2 steps forward, 1 step back approach was draining as hell. And according to a CRW guy I talked to at the visitor’s center, much of it could have been avoided. He recommended coming from the north instead of through Princeton. Apparently much of the up down on the south approach is replaced with a single steep climb, called Mile Hill Road. 1 mile, 9% grade. And he had tips for the south approach as well, to avoid a few of the initial climbs. Useful info, should I ever do this ride again
I got to the visitor’s center with no real idea how much more climbing was left. I figured I had a thousand feet or so, but luckily it was only about 700. And the final climb up to the summit wasn’t really all that steep. All the same, there was no way I was going to be able to reverse directions and ride back to Cambridge after all that. For one thing I didn’t have a usable route after Acton, which meant going back to asking people for directions. And for another, my legs were just cooked. It only took me 4 hours of riding to do the 61 miles and climb to the summit. Definitely need to ride at an easier pace if I want to turn that into 120 miles.

Soo, my pal Vieve came and picked me up. Was really really nice of her, considering it was an hour each way by car. I owe her at least another dinner. Sitting in the passenger seat (and later standing in a hot shower) I realized just how cold I’d been all day. A bike jersey, bib shorts, and arm warmers are not enough on a day where the high is 60.

Ride info here. check out that elevation profile. gah.

update: Clearly i need to do more research before leaving on such long rides. Just found this route on the CRW website.