-*-outline-*-

* Ensure it plays nicely with tramp.

  After loading both tramp and cygwin-mount, when I do C-x d /tmp RET,
  I see

       /:
       wildcard tmp
     (No match)
     total 0

  But if I remove the tramp handler from some hook or other (I forget
  which), cygwin-mount resumes working perfectly.  (My current
  workaround is to avoid loading tramp if cygwin-mount is already
  loaded.)

* Write lots of tests

* Make `unload-feature' Do The Right Thing
   namely remove stuff from file-name-handler-alist.  This appears not
   to happen automagically, so we probably need to define
   cygwin-mount-unload-hook.

* lazy initialization
   Currently, the mount table gets filled in when you run
   cygwin-mount-activate.  However, it might be cleaner to have
   cygwin-mount-substitute-longest-mount-name fill the table in if
   it's empty.  Ideally it'd be nice to entirely eliminate
   cygwin-mount-activate (and cygwin-mount-deactivate).

* Consider making a Cygwin package of cygwin-mount
   should it conflict with Cygwin Emacs?  Does Cygwin's setup even
   *have* the concept of conflict?  Or perhaps just wrap the whole
   thing with, in effect, "load this file only if we're Win32 and not
   cygwin".

* Test customization.
  Of course I have almost no idea how customization works :(

* Consider flipping the cells in the cygwin-mount-table alist.
  That is, put the cygwin directory in the car, and the "windows
  device" in the CDR.  This won't make things work any better, but
  it'll conform to standard Lisp practice of having the key in the
  car, and the sought data in the CDR.  At least, I *think* the cygwin
  directory is the key, and the "windows device" is the sought data.

* Closely related to the above: write unit tests for the function 
  cygwin-mount-substitute-longest-mount-name.  That function appears
  to be the heart of the whole matter, and it looks like it'd be easy
  to test -- there's one free variable (namely,
  cygwin-mount-table--internal, the alist referred to above), and one
  input parameter.