Every time I opened the blog locally to work on it, every page crashed. Nothing wrong on the live site — this only happened on my computer, in development mode. But still.
What happened
The framework does a behind-the-scenes process each time a page loads locally. Part of that process copies some page settings — including the site's URL. There's a quirk: if the URL is stored as a special URL object instead of a plain text string, the copy process chokes on it.
The fix was simple: use a plain text string for the URL. Same result, no crash.
Eight lines changed. The hour and a half I spent before that adding debug logging to figure out which setting was causing it — that doesn't count.