Lost ability to navigate to parent directory

[jcmuller]

How can I begin to debug this issue?

[palikar]

The button for the parent directory is inserted in the function neo-buffer--insert-root-entry. Maybe you can start form there. Also, what do you mean exactly? Did you do something with your config and broke something or did you played with the code of neotree itself?

[mrspeaker]

I think I have this issue too: if you use n/p to navigate to the line for the parent directory ("> ..") then press SPC or RET - nothing happens. I expect it to move to the parent directory.

To actually move to the parent directory you need to move your cursor over until it is on the ".." and THEN press spc or ret (or use the shortcut key).

This is different behaviour than all other files and directory where they can be activated from anywhere on the line.

[mrspeaker]

I did a little digging into this and think I can see where the issue is, but I'm not sure the correct way to fix it. Currently the parent directory line ".." is treated the same as every other line, and when you hit RET it tries to unfold it but because it's above the current path it does nothing.

As a "proof of concept" I made a dodgy hack by changing neo-open-dir from this:

(defun neo-open-dir (full-path &optional arg)
  (if neo-click-changes-root
      (neotree-change-root)
    (let ((new-state (neo-buffer--toggle-expand full-path)))
      ... stuff to expand the path

To a version where there is an "or" check to also change root if full-path is a parent directory:

(defun neo-open-dir (full-path &optional arg)
  (if (or neo-click-changes-root
          (< (length full-path) (length neo-buffer--start-node)))
      (neotree-change-root)
    (let ((new-state (neo-buffer--toggle-expand full-path)))
      ... etc

However, I guess this is a bad way to do it - it shouldn't be based on the path string length: it should actually know it's the parent! The code above does work (and hitting SPC or RET on ".." will take you up a directory), but I suspect it will do weird things on symlinks. Maybe.

Anyway, if I figure out a more robust way to do it I'll make a pull request.

[EDIT: I think this is a better way, that actually checks the nodes not just string length:

 (if (or neo-click-changes-root
          (equal full-path (neo-path--updir neo-buffer--start-node)))
      (neotree-change-root)

]