miniblog.

I had a third attempt at porting difftastic from Dijkstra to A*, and I've eventually concluded that A* just isn't worth it for the implementation complexity. I had fun writing it up though, and I've even attempted ASCII art for building intuitions:

I'm comfortable with git, and recently learned mercurial, and I've concluded: commits are the wrong abstraction. Work-in-progress and finished work are different concepts. Rebases and even branches are for WIP. Signing is for finished work. Timestamps and authors change for WIP.

What's the best poison value to use for kernel malloc? OpenBSD eventually concluded that it depends on the arch: