Fix #394: Avoid problem happening.

[starcraftman]

This is largely a response to fix: #394.

Maps are only bound on prepare when function is guaranteed synchronous, else delay until finish of update/install to be done in s:finish(). Avoids the accidental neovim breakage experienced in that issue.

Also, small typo with respect to the platform test, print doesn't need brackets for python 2.

TODO:

• Don't write pending updates if no commit/tag opts.
• Resolve buffer reuse.

[junegunn]

Thank you. The remaining problem is as we continuously reuse the buffer when it's open, it fails when you do PlugUpdate again while the window is open and press D. We can either 1. unmap those mappings, 2. or change the code not to reuse the buffer but simply wipes it out and create another one (enew?).

[starcraftman]

@junegunn Good point, I'll add that to my todo.

[starcraftman]

@junegunn Anything else? I went with simply destroying it every time by invoking q.

[junegunn]

How about enew?

  if s:switch_in()
    enew
  else
    call s:new_window()
  endif

This clears the buffer (bufhidden=wipe) and its maps but reuses the window, so the size of the window is retained if the user previously resized it for some reason.

[junegunn]

After a second thought, I think q (which is essentially bd) is safer. A user might tab split the buffer :)

[junegunn]

👍

[starcraftman]

No problem.