On Sun, Nov 06 2022, Đoàn Trần Công Danh wrote: > In a later change, we will convert the bisect--helper to be builtin > bisect. Let's start by pretending it's the real bisect when reporting > error. It's good to have this in here, FWIW I left this out of my version reasoning that it was broken before, so we could leave it with the psot-cleanups. I'm fine with having it in while we're at it, but just FWIW: I don't think we're "pretending". We *are* the bisect command, the fact that we're implementing it via a "helper" is really quite irrelevant, just like we're not "pretending" if we say we're "git bisect" instead of "git-bisect.sh" or whatever. The only cases where we should have these helpers report that they are the helper is for messages that aren't user-visible, e.g. the "-h" output emitted when you invoke a helper-specific command that no user should be invoking (although arguably, we should just BUG() out there...).