Question

I’ve tried to write a simple test (kill-buffer-test.el):

:; exec emacs -Q --script "$0" -- "$@"

(setq debug-on-error t)

(setq n 0)
(dolist (idx (buffer-list))
  (message "%d '%s' %s" n idx (buffer-list))
  (setq n (+ 1 n))
  (message "result: %s cb='%s'\n" (kill-buffer idx) (current-buffer)))

And run it:

$ ./kill-buffer-test.el
0 '*scratch*' (*scratch*  *Minibuf-0* *Messages*  *code-conversion-work*  *load*)
result: t cb='*Messages*'

1 ' *Minibuf-0*' ( *Minibuf-0* *Messages*  *code-conversion-work*  *load*)
result: nil cb='*Messages*'

2 '*Messages*' ( *Minibuf-0* *Messages*  *code-conversion-work*  *load*)
result: t cb='*scratch*'

3 ' *code-conversion-work*' ( *Minibuf-0*  *code-conversion-work*  *load* *scratch* *Messages*)
result: t cb='*scratch*'

4 ' *load*' ( *Minibuf-0*  *load* *scratch* *Messages*)
result: t cb='*scratch*'

Selecting deleted buffer
$ echo $?
255

Notice that *Minibuf-0* buffer wasn’t killed (why?), *scratch* was resurrected, and emacs exited with a strange error message.

So, should I worry that (current-buffer) may fail? And if I should, how to test that? (current-buffer) will raise an error or will just return nil?

M-x version
GNU Emacs 24.3.50.1 (i686-pc-linux-gnu, X toolkit) of 2013-02-08 on linux.9bf016

Answer 1

No need to worry: (current-buffer) can’t fail (Emacs would crash and burn if it ever did).
*scratch* as well as a few other internal buffers get re-generated as/when needed.