Re: emacs-25 ed1b2de: Fix NNTP NEWGROUPS off-by-a-few-hours bug

Re: emacs-25 ed1b2de: Fix NNTP NEWGROUPS off-by-a-few-hours bug

[Katsumi Yamaoka]

Hi,

On Wed, 13 Jan 2016 16:29:37 +0000, Paul Eggert <eggert@cs.ucla.edu> wrote:
> branch: emacs-25
> commit ed1b2de50ae87d8b2f9882be7576fbe5097df65b
> Author: Paul Eggert <eggert@cs.ucla.edu>
> Commit: Paul Eggert <eggert@cs.ucla.edu>

>     Fix NNTP NEWGROUPS off-by-a-few-hours bug

>     * lisp/gnus/nntp.el (nntp-request-newgroups): Format string
>     in Universal Time, since we’re telling the server “GMT”.
> ---
>  lisp/gnus/nntp.el |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)

> diff --git a/lisp/gnus/nntp.el b/lisp/gnus/nntp.el
> index 0e10dfd..0006ef9 100644
> --- a/lisp/gnus/nntp.el
> +++ b/lisp/gnus/nntp.el
> @@ -1130,7 +1130,7 @@ command whose response triggered the error."
>         (prog1
>             (nntp-send-command
>              "^\\.\r?\n" "NEWGROUPS"
> -            (format-time-string "%y%m%d %H%M%S" time)
> +            (format-time-string "%y%m%d %H%M%S" time t)
>              "GMT")
>           (nntp-decode-text))))))

Is this really a fix?  In Japan (i.e., +0900) this doesn't seem
to pass the correct GMT date to a server:

(defun nntp-request-newgroups-test (date)
  (let* ((time (date-to-time date))
	 (ls (- (cadr time) (nth 8 (decode-time time)))))
    (cond ((< ls 0)
	   (setcar time (1- (car time)))
	   (setcar (cdr time) (+ ls 65536)))
	  ((>= ls 65536)
	   (setcar time (1+ (car time)))
	   (setcar (cdr time) (- ls 65536)))
	  (t
	   (setcar (cdr time) ls)))
    (format-time-string "%y%m%d %H%M%S" time t)))

(let ((date (message-make-date)))
  (list date
	(nntp-request-newgroups-test date)))
 => ("Thu, 14 Jan 2016 18:44:10 +0900" "160114 004410")

If removing the last `t', it returns:

 => ("Thu, 14 Jan 2016 18:44:18 +0900" "160114 094418")

See also `gnus-ask-server-for-new-groups'.

Regards,

Re: emacs-25 ed1b2de: Fix NNTP NEWGROUPS off-by-a-few-hours bug

[Andreas Schwab]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> Is this really a fix?  In Japan (i.e., +0900) this doesn't seem
> to pass the correct GMT date to a server:
>
> (defun nntp-request-newgroups-test (date)
>   (let* ((time (date-to-time date))
> 	 (ls (- (cadr time) (nth 8 (decode-time time)))))
>     (cond ((< ls 0)
> 	   (setcar time (1- (car time)))
> 	   (setcar (cdr time) (+ ls 65536)))
> 	  ((>= ls 65536)
> 	   (setcar time (1+ (car time)))
> 	   (setcar (cdr time) (- ls 65536)))
> 	  (t
> 	   (setcar (cdr time) ls)))
>     (format-time-string "%y%m%d %H%M%S" time t)))
>
> (let ((date (message-make-date)))
>   (list date
> 	(nntp-request-newgroups-test date)))
>  => ("Thu, 14 Jan 2016 18:44:10 +0900" "160114 004410")
>
> If removing the last `t', it returns:
>
>  => ("Thu, 14 Jan 2016 18:44:18 +0900" "160114 094418")

The manual timezone adjustment needs to be removed as well.

Andreas.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."

Re: emacs-25 ed1b2de: Fix NNTP NEWGROUPS off-by-a-few-hours bug

[Katsumi Yamaoka]

On Thu, 14 Jan 2016 11:15:13 +0100, Andreas Schwab wrote:
> The manual timezone adjustment needs to be removed as well.

Ah, thanks.  Committed.

Re: emacs-25 ed1b2de: Fix NNTP NEWGROUPS off-by-a-few-hours bug

[Paul Eggert]

On 01/14/2016 03:26 AM, Katsumi Yamaoka wrote:
> Ah, thanks.  Committed.

Thanks for fixing my thinko; I should have noticed the nearby timezone 
mumbo jumbo.