After the all the noise surrounding the forking of the Zig compiler in order to speed up builds, I'm really surprised that this isn't the top comment. The fact that a one-man team achieved 1s build times proves pretty conclusively that slow build times were entirely result of negligent development practices, and that the fork was a complete misallocation of time.
I'm pretty sure they forked Zig long before it supported incremental compilation. In fact, it's still experimental and not enabled by default.
A significant part of the port was also for the built-in safety. Unsafe code just... fails to compile.
Turns out there's a journey here.