Hello, Dale Sedivec <d...@codefu.org> writes:
> Hi! I think clocktables may be parsing time zones incorrectly in :tstart > and :tend, leading to incorrect output. Steps to reproduce: > > 1. Start a fresh Emacs within that same along the lines of: > > ~~~~~~ > mkdir /tmp/emacs > TZ=America/Chicago HOME=/tmp/emacs /path/to/emacs > ~~~~~~ > > I *think* setting TZ to a non-UTC time zone is important to reproduce this. > > 2. Install latest org-mode from Git. I did this via package-install-file > on a package I built from a Git checkout of the org-mode repository. > > 3. Open an org-mode buffer, e.g. C-x C-f ~/test.org RET. > > 4. Insert: > > ~~~~~~ > > #+BEGIN: clocktable :maxlevel 2 :scope file :step day :tstart "<2017-07-02 > Sun>" :tend "<2017-07-09 Sun>" > > #+END: > ~~~~~~ > > 5. Save the buffer. > > 6. Move point to the start of that dblock and C-c C-c to update it. > > Expected results: First step is "Daily report: [2017-07-02 Sun]" which > matches the :tstart. > > Actual results: First step is "Daily report: [2017-07-01 Sat]", the day > before the :tstart date. The last step is also 2017-07-07, not 2017-07-08 > as I was expecting. > > I notice that as of 112c5ba479d, org-clocktable-steps parses :tstart and > :tend with the ZONE argument to org-parse-time-string as T. I think this > is causing org-parse-time-string to parse these user-entered dates as UTC > rather than the user's local time as I would have expected. Changing > org-clocktable-steps from doing (org-parse-time-string ts nil t) > to (org-parse-time-string ts nil nil), and then the same for te as well, > seems to fix this problem. Fixed. Thank you for the report and the analysis. Regards, -- Nicolas Goaziou
