One thing I've seen smart developers do on multiple occasions is unknowingly throw out their ability to reason about their code while adding caching ( https://softwareengineering.stackexchange.com/questions/3512... ). Then when some weird bugs start showing up it's like "Ummm... this shouldn't be happening" and then end up having to restart _everything_ because the interplay between layers of caching and the rest of the program no longer make any sense.
> So at least on Linux, cache performance hasn’t been an issue with LMDB. It’s noticeably worse on Windows, as is memory management in general on Windows, but obviously Windows is not a high performance platform to begin with.
This made me smile.
It's also why "just run WSL" or "just run Linux in a VM on OSX" is unsatisfying. The performance characteristics are quite different, which makes a big difference if you're trying to reproduce a race condition.
What the hell is going on with the line breaks there? That's painful to read on mobile
Wordpress editor being wonky. Should be better now.