Re: ordering candidates when using completion-at-point-functions

[Dirk-Jan C . Binnema]

Hi Stefan,

On Thu, Jul 04 2013, monnier@IRO.UMontreal.CA wrote:

>> Instead, I would like to deliver the candidates in order of frequency; I
>
> What do you mean by "deliver"?

Well, I mean the order in which the candidates are presented in the
*Completions* buffers, as well as the order in which in can cycle
through them.

>> (An alternative might be to use a display-sort-function; it seems the
>
> Using display-sort-function sounds about right, yes.

>> current org-contacts[2] does something like that; but this seems /very/
>> complicated solution, which is hopefully not necessary for my modest
>> needs...)
>
> Which part do you find complicated?  We can probably provide some helper
> function to make it simpler.

So, boiled down to the essentials, I have something like this for my
message-composition buffer:

----
;; already sorted
(setq candidates'("foo1@example.com" "fnorb2@example.com"
                   "cuux3@example.com" "bar4example.com"))

(defun my-completion-function (&optional start)
  (let ((end (point))
         (start (or start
                  (save-excursion
                    (re-search-backward "\\(\\`\\|[\n:,]\\)[ \t]*")
                    (goto-char (match-end 0))
                    (point)))))
    (list start (point) candidates)))
 
(add-to-list 'completion-at-point-functions 'my-completion-function)
---

So, completing after e.g. "To: ", I get:

----
Possible completions are:
bar4@example.com            cuux2@example.com
fnorb3@example.com          foo1@example.com
----

That is, the results are ordered alphabetically, rather than the 1-2-3-4
order I would like, as in my already-sorted list.

So, my question is how I can influence the sorting order -- either by
telling the completion machinery to not try to sort my candidates, or,
if that's not possible, to provide a display-sort-function and/or
cycle-sort-function.

I tried to write a completion function as per
   
https://www.gnu.org/software/emacs/manual/html_node/elisp/Programmed-Completion.html
(and org-contacts, `org-contacts-make-collection-prefix'), but that
seems quite a bit of complexity, just to set the sort-functions to
identity. Maybe I'm missing something obvious -- or would that be the
right way to go forward anyway?

Kind regards,
Dirk.

-- 
Dirk-Jan C. Binnema                  Helsinki, Finland
e:djcb@djcbsoftware.nl           w:www.djcbsoftware.nl
pgp: D09C E664 897D 7D39 5047 A178 E96A C7A1 017D DA3C