Re: [O] Problem with org-timestamp-up and timezones

[Derek Feichtinger]

Hi

Robert Eckl <eckl.r <at> gmx.de> writes:

> > 8.3.4 / 8.3.4-15-gdd9be3-elpaplus). So this must be an incompatibility
> > introduced in the emacs core development.
> > 
> The bug seems to be introduced after emacs commit c23c965bb9d0
> 

Thanks for the pointer. I can see that something with the timezones was
changed, but cursorily looking just showed me some changes in the comments.
I only had time now to have a closer look. But I decided to try fix it in
org mode, since this seemed easier.

The problematic statement is in org.el, defun org-timestamp-change, where
the argument list to encode time evaluates to something like

(encode-time 0 0 0 30 11 2013 '(nil nil nil))

i.e. the 7th argument is a list, while the function would be ok with getting
the three nil values not packed into a list. So, just introducing an "apply"
in front fixes this for now.

So from the following code in org-timestamp-change

   ...
      (setq time
            (encode-time (or (car time0) 0)
             (+ (if (eq org-ts-what 'minute) n 0) (nth 1 time0))
             (+ (if (eq org-ts-what 'hour) n 0)   (nth 2 time0))
             (+ (if (eq org-ts-what 'day) n 0)    (nth 3 time0))
             (+ (if (eq org-ts-what 'month) n 0)  (nth 4 time0))
             (+ (if (eq org-ts-what 'year) n 0)   (nth 5 time0))
             (nthcdr 6 time0)))
   ...

I just modify to the following

   ...
      (setq time
            (apply
             'encode-time (or (car time0) 0)
             (+ (if (eq org-ts-what 'minute) n 0) (nth 1 time0))
             (+ (if (eq org-ts-what 'hour) n 0)   (nth 2 time0))
             (+ (if (eq org-ts-what 'day) n 0)    (nth 3 time0))
             (+ (if (eq org-ts-what 'month) n 0)  (nth 4 time0))
             (+ (if (eq org-ts-what 'year) n 0)   (nth 5 time0))
             (nthcdr 6 time0)))
   ...


Afterwards the time shifts on the clock lines works fine, again. But I do
not know whether this fixes all cases, or what really is the deeper reason
that this error surfaced right now.

Cheers,
Derek