I was updating my emacs backup approach using ideas from answers to this question, but I'm left unsure by one aspect, namely: the fact that several of them arrange to have backups placed in a single backup directory. So they might have something like:
(setq backup-directory-alist `(("." . "~/.emacs.d/backups")))
That's in contrast to what I've been doing up until now, with each "actual" directory having it's own backup directory, using something like:
(setq backup-directory-alist `(("." . "backups")))
That has seemed sensible to me since it keeps any given file's backups close to it in the filesystem. As a result, if and when I need a copy of a backup, I just drop into ./backups
instead of having to climb out of whatever Mines of Moria-esque subdir I've been working in, drag my sorry a*se all the way over to ~/.emacs.d/backups
or the like, and then dive into a haystack of mostly unrelated files with blindingly !-infested full pathnames, to find the needle that is my desired backup. Not to mention the fact that backing up to a single directory creates a risk that sensitive files may get backed up to a place with less than sensitive permissions.
Nevertheless, reading around the topic I now get the impression that the single directory approach may in fact be the norm, and that I'm the weirdo.
So what am I missing? Is there some gotcha to the per-directory approach? Or some really useful benefit with the single directory approach? Or both? Or what?