Uwe Brauer <oub@mat.ucm.es> writes:

> I tried 
>
> (setq gnus-thread-sort-functions
>       '(not gnus-thread-sort-by-most-recent-date))
> ;;        '(not gnus-thread-sort-by-date))
>
> (setq gnus-sort-functions
>       '(not gnus-article-sort-by-most-recent-date))
> ;;        '(not gnus-article-sort-by-date))
>
>
> Or 
> (setq gnus-thread-sort-functions
>         '(not gnus-thread-sort-by-date))
>
> (setq gnus-sort-functions
>         '(not gnus-article-sort-by-date))

Those are all invalid.  The variables' values must all be lists, and in
order to negate an entry, you'd turn it into a list with car equal to
not, e.g., it should be:

(setq gnus-thread-sort-functions
      '((not gnus-thread-sort-by-number)
	(not gnus-thread-sort-by-most-recent-date)))

> And not as it should be 
>
> []{ }O  57606 [  Uwe Braue] Uwe Brauer  [[CV] (was: [Amnu 19])    ] Sun, 28 Apr 2019 11:30:41 +0200 8.8k 6   
> []{ }OA     57609 [ +henar de ] Uwe Brauer  [                         ] Sun, 28 Apr 2019 09:35:45 +0000 10k 13   
> []{ }O  57585 [  tex-d-l-r] tex-d-l@li  [tex-d-l Nachrichtenkompil] Sun, 28 Apr 2019 00:00:07 +0200 13k 165   
> []{ }O  57593 [  users    ] oub@mat.uc  [Subject: Digest of users@] Sun, 28 Apr 2019 08:00:27 +0000 (UTC) 38k 11   
> []{ }O  57621 [  Uwe Braue] Uwe Brauer  [[stuff] (was: [CV])      ] Sun, 28 Apr 2019 16:04:34 +0200 9.4k 16   

Sorry, I don't see how the sorting is here.  Converted to UTC, the times
here are

- 13:30:41
- 09:35:45
- 02:00:07
- 08:00:27
- 18:04:34

and they are not ordered monotonically.  So which criterion do you
actually want to use for sorting.

FWIW, that's what I use.

--8<---------------cut here---------------start------------->8---
(setq gnus-thread-sort-functions '(gnus-thread-sort-by-number
				   gnus-thread-sort-by-most-recent-date
				   gnus-thread-sort-by-total-score)
      gnus-subthread-sort-functions '(gnus-thread-sort-by-number
				      gnus-article-sort-by-date))
--8<---------------cut here---------------end--------------->8---


This gives me high-score threads on top, and when the thread scores are
equal, the thread with a newer message is above the other.

Inside a thread, e.g., when there are multiple replies to the very same
article, I sort chronologically with older above newer messages (so that
I don't read the second reply before the first).

Bye,
Tassilo