Read it as `ln (-s x) y`, not `(ln -s) (x y)` I could never remember the operand order for `ln -s x y`, and now I've realized why: the command supports two simultaneous parsings. `(ln -s) (x y)` — the intended reading. `-s` for "symbolic," argument order same as in `cp x y`. Fine, but I don't trust such analogies — not after `find`, `dd`, or `tar`. Also, it is weird how at birth we denote the symlink as `x y`, but later if we `ls -l y` we'll see `y -> x`. Why the reversal? Using `ln -s` makes `-s` powerless to impose a convention: only the link itself is qualified as symbolic, and it is left to us to figure out what that means for the operands. `ln (-s x) y` — my reading. `-s` for "source." You're declaring x as the source of content for the new name y. "But wait, x is called the 'target' in symlink terminology!" This was my confusion. I'd been treating "source" and "target" as antonyms, so the mnemonic kept breaking. But x is both: target of the link, source of the content.¹ All symlinks to a resource form a tree rooted at the original: v1/ ← original tl;dr: `ln -s old new` pushes `new` onto a stack rooted at `old`. The `-s` is for "source," not just "symbolic." --- ¹ Like how topology students eventually realize a set can be both closed and open — the words aren't antonyms, just independent properties. I wonder what formal topology scaffolding could make "source" and "target" correspond to "open" and "closed." |