7 min
Watch it fail first
The same bug arrived twice, 18 months apart, on the same store, and the second time I found my own fix from the first time sitting in the git history with nothing next to it that could have caught the repeat.
Debugging
All writing →8 posts filed under Debugging, newest first.
8 min
Rubber duck debugging, in writing
Half the questions I write in a channel never get sent, because I figure out the answer somewhere in the middle of writing them, and I've stopped thinking of that as wasted effort.
9 min
z-index: 9999
A size dropdown on a client's product grid kept rendering underneath the row of products below it, and the 9999 already sitting on it had been doing nothing at all for 3 years.
8 min
I asked them to hard-refresh
Two customers wrote in to say the sale wasn't on the store's homepage, I opened the homepage and the sale was right there, and it took me most of an afternoon to figure out that the difference between us was me.
8 min
Flaky tests usually aren't
A checkout test failed about 1 run in 15, everybody called it flaky and re-ran the job for months, and the whole time it was reporting a real race that customers were hitting too.
7 min
position: sticky and its containing block
Four declarations, all of them correct, and the one line stopping them from working was something I wrote three years ago on an element seven ancestors up.
2 min
git bisect
Halving a commit log is a mechanical process that finds a regression while you make coffee, and the only genuinely hard part is writing down what you are looking for before you start.
6 min
View source
The web taught a generation of us by being readable and I'm not sure the thing replacing it has that property, which might matter more than it sounds like it does.