Re: docstring of display-completion-list

[Roland Winkler]

kevinr@ihs.com (Kevin Rodgers) writes:

> [Followup-To: gnu.emacs.help]
> 
> Roland Winkler wrote:
> > 
> > Attached below is the code of a function my-complete-symbol. Most of
> > the code I borrowed from lisp-complete-symbol. The function works
> > completely as expected. However, according to the docstring of
> > display-completion-list used at the end of this function there is
> > one thing I do not understand at all:
> > 
> > If I choose a completion from the `*Completions*' buffer, why is the
> > text inserted in the buffer from where I called my-complete-symbol?
> > 
> > According to the docstring of display-completion-list it runs the
> > hook completion-setup-hook which (I think by default) calls the
> > undocumented function completion-setup-function. Is the latter
> > inserting the completion in the buffer from where I called
> > my-complete-symbol?
> 
> No, I think it's this bit in my-complete-symbol itself:
> 
> >           ((not (string= pattern completion))
> >            (delete-region beg end)
> >            (insert completion))

No, I don't think so. This cond clause is evaluated when pattern is
shorter than completion so that inserting something makes sense. The
last cond clause with the call to display-completion-list is
evaluated when pattern equals completion and one really has to make
a choice from the *Completion* buffer. You can check that the above
clause is not evaluated when the completion buffer is generated and
one makes a choice in it. So what's going on here??

Roland