• (migrated from [ospaths] BUG: splitFile("/a.txt").dir = "" ; + other bugs with splitFile #8255):

• bug 2:

• parentDir("a/b//")==a/b is a bug
  (but parentDir("a/b/") == a is OK)

EDIT: fixed

• bug 3:
  echo parentDir("/")
  empty, which is a bug
  rdmd --eval 'std.path.dirName("/").writeln'
  /
  os.path.split("/")
  ('/', '')
  Curiously, echo splitFile("/") is correct:
  (dir: "/", name: "", ext: "")
  shouldn't parentDir be implemented in terms of splitFile? or is that for efficiency optimization ? (if so, see [WIP] nim needs a slice type analag to D's T[] dynamic array #8256)

EDIT: still not fixed; whichever convention is used, splitFile vs parentDir still disagree

• bug 4:
  echo parentDir("/a.txt")
  empty, which is a bug (and inconsistent with splitFile)

EDIT: still not fixed

• bug 5
  splitFile documentation says: "dir does not end in DirSep" but should say "dir does not end in DirSep" unless it's "/"
  EDIT: still not fixed

• bug 6 [NEW]
  parentDir("foo") returns "" instead of "."

NOTE: D and python differ for the following case:

D: rdmd --eval 'std.path.dirName("/a/a/").writeln' /a python: os.path.split("/a/a/") ('/a/a', '') nim: echo parentDir("/a/a/") /a 

that one should be documented

• bug 7:
  parentDir(".") is empty instead of ".."
  parentDir("..") is empty instead of "../.."