|
|
|
unexpected race condition
Wednesday, May 27 2026
The other day I looked at the east part of the Chamomile Wall, the part I'd built mostly in 2019 in something of a manic disorganized hurry over the course of numerous short after-work bursts. At the time I'd striven mostly for height over anything else, so the wall was initally thin and shoddy, with numerous places where one could see daylight between the rocks from the other side. Over time, though, as partial collapses needed to be repaired, I've bulked up the wall over most of its length, making it both thicker and more substantial. But it is still subject to frost heave, and in places is still so thin that such movement causes parts of the wall to pull away from each other. The part I was looking at was about fifteen feet from wall's eastern end and it was clear that spaces had opened up between stones. In a few years the spaces would be wide enough that the wall would be in danger of collapse. As always, the solution for such issues is to bulk up the wall from below on the north (downhill) side. This afternoon I built a primitive buttress consisting of two stacked piles of irregularly-shaped stones about eighteen inches apart. I then spanned the distance beween them with a curved stick of bluestone (curved-side-up of course) and then lay down a layer of flat stones atop that with their lower parts against the wall and their higher parts atop this buttress. From there, I was able to continue stacking stones nearly all the way to the top of the wall, making it much thicker and more solid in that previously-weak location.
This evening while Gretchen was off at guitar practice and pilates, I tried to get Charlotte to join me on a walk down the Stick Trail, but she didn't end up coming, perhaps because I was carrying the big Kobalt chainsaw, and she knew it would be boring if I was going to be cutting wood. I used that saw to completely clear the short connector trail between the Gullies Trail and Chamomile Gorge Trail, a section normally only used by Crazy Dave and his dogs. The hope is that he will see how easy the hiking is that way now and switch over to the Gullies Trail and I'll encounter him less on the Stick Trail. Neither he nor I like encountering other humans in the forest, but the main issue here is Charlotte's beef with Dave's dog Brigette. Whenever they encounter each other, it's an unplesant situation.
I'm also hoping to get started on clearing a path down the creek bed of the Chamomile itself, though I didn't start on that today because it would be a big project and it was getting towards the hours where Dave and his dogs venture out on the trail system.
This evening while toggling into and out of fast communication mode on my local test laboratory ESP8266, I decided to add a "dump pin state" command. In the process of getting that to work, I realized that the normally rock-solid web page that displayed the device_feature (pin) information for a device was getting confused about what state the pins were in, though only when fast communication mode was on. Something about having the ESP8266 communicate both via polling and via websocket was slowing something down enough that a race condition had opened up. This was likely aggravated by the fact that the web page in question has an auto-update feature, periodically changing the data it displays to reflect changes in the data in the underlying database. Not only that, but changes made to pin-state checkboxes in the tabular display are sent to the device with the next poll, meaning there can occasionally be confusion between the auto-update process (which also updates the states of checkboxes!) and the process that responds to user changes to checkboxes by updating the database. In descibing this, it sure sounds like a recipe for race conditions, but, prior to the development of the new websocket-based communication channel, this hadn't been a problem.
For linking purposes this article's URL is:
http://asecular.com/blog.php?260527 feedback
previous | the future |
