|
|
what people with dashboard cams listen to Friday, October 26 2018
At work today, the first task I set for myself was to debug a weird problem with one of our sites. The site ran on a static Javascript Ember frontend with a backend written in Ruby on Rails. That backend, in turn, stored its data on a Azure database in Microsoft's cloud (a me-too technology I approach with skepticism). The problem was that the Azure database had to be moved to a new account to separate it completely from a departing employee (John the developer who left earlier this week) but for some reason the new Azure database kept crashing the Rails application. By crash, I mean a bunch of errors and, if it crashed twice in a row, an actual segment fault. The messages from these crashes were cryptic and unhelpful, making my already bad predisposition towards Ruby on Rails even more negative (and also damaging my feelings about Azure at the same time). Since the errors were so vague, the problem was very difficult to debug. I managed to figure out how to swap database connection parameters in an irb session so I could experiment with different ones (just to see how, for example, it failed with bad credentials). At one point I was so confused by all the groups of connection parameters that I fooled myself into thinking I'd fixed the problem. It was unfortunate that I sent an email to my boss and head honcho prematurely declaring the problem solved. But I walked that back as soon as it was clear I was wrong.
For lunch, I had a pair of frozen vegetable burritos from Trader Joe's. They weren't very good.
This afternoon, I faced another daunting debugging issue when a script that had worked fine for one customer completely blew up on another. What was going wrong? Was I losing my mind? After initially thinking the database permissions had been changed by Morning Dave, I realized that there was no way my T-SQL BULK INSERTS were ever going to work unless the database was on the same server as the Python script that processed the SQL script. That meant that the idea of putting an intermediate database on a different server was mistaken, not my script (I hadn't remembered how I'd set all this up only a month and a half ago). Once I had that all figured, I climbed into my Subaru and began my weekend. As I neared the Route 199 stoplight on 9W northbound, the check engine light blinked on. It was a P0139 error, meaning I had to lower the value of the 47 microfarad capacitor I'd installed last night. My next value would be 22 microfarads. If that produces a P0420 error, I'll raise it to 30 microfarads. If it produces a P0139 error, I'll try something like 10 microfarads.
Back at the house, I quickly gathered a backpack of fairly dry firewood from just south of the Chamomile on the Stick Trail. Then I took 150 milligrams of diphenhydramine and climbed into bed with my laptop and watched YouTube channels featuring endless depictions of car crashes in the United States (in the past, I'd only seen such collections made in Russia). Part of what makes these videos interesting is the random things you hear the drivers listening to just before and after the crashes recorded by their dashboard cameras. Some people are listening to pop music, some rap. A lot of people seem to be listening to books being read to them (what we used to call "books on tape"). What you don't hear much of is country music or rock 'n' roll. Evidently dashboard cams haven't penetrated deeply into those demographics.
For linking purposes this article's URL is: http://asecular.com/blog.php?181026 feedback previous | next |