I've been using the Emacs timeclock to track my time on various
activities. When I invoke timeclock-in, I'm unable to enter a project
name containing embedded space characters (unless it's already in the
auto complete list by virtue of me manually editing the timelog file
beforehand).

I think I tracked the problem down to the completing-read function
that eventually gets invoked from timeclock-in. I put the following
snippet (from timeclock.el) in the *scratch* buffer and invoked it via
C-M-x

(completing-read "Enter string (default foo): " '("bar" "fuz") nil nil
nil nil "foo")

If I type b the minibuffer will complete to bar, but I can't
then type a space. My temporary work around was to do the following:

(global-set-key (kbd "C-#") (lambda () (interactive) (insert-char ?\s
1)))

and then press C-# instead of while in the minibuffer, and
that works fine :)

Brian Adkins gmail.com> writes:

C-q SPC

On Mar 7, 4:26 pm, Pascal Bourguignon informatimago.com> wrote:

Thanks, that's a much simpler work around than mine :)

That would work for most of my timeclock projects (0 to 2 spaces), but
occasionally I just want to enter a longer description on that line,
such as:

Met with John & Joe to discuss the authentication scheme

so, it would also be nice to get my space bar back from the over
zealous auto completer. I can M-x timeclock-visit-timelog and edit the
line manually, but it would be nicer to just enter it in the
minibuffer.

Sure, you can quote each space character with `C-q'. Or you can fix Emacs to
do what you want all the time.

Emacs users should be able to type space chars in minibuffer input. Period.
Emacs 22 finally allows that for file-name input, but it should be available
for all minibuffer input. Input completion is not limited to command names.

Solutions:

1. Use Icicles: http://www.emacswiki.org/cgi-bin/wiki/Icicles.

2. If you don't want to use Icicles, but you do want to allow space chars in
minibuffer input, do this:

(define-key minibuffer-local-completion-map
" " 'self-insert-command))
(define-key minibuffer-local-must-match-map
" " 'self-insert-command))

On Mar 7, 5:09 pm, "Drew Adams" oracle.com> wrote:

I think you mean to set those keys to nil rather than to
`self-insert-command'. It may have the same result in 99%% of the cases
in the end, of course.

Stefan

Right. It's enough to remove the local binding, to restore the global
binding. Either is OK here, but what you suggest is preferable.