* lisp/org-clock.el (org-clock-special-range): Use nil to represent
`untilnow'.
* lisp/org-clock.el (org-clocktable-steps): For `untilnow' block, set
set timestamp to 2003.
* doc/org-manual.org (The clock table): Document `untilnow' floor when
used with :step.
For `untilnow', org-clock-special-range sets the start to
"<-50001-11-30 Tue 00:00>", but org-parse-time-string actually assumes
a YYYY-MM-DD format and parses the year as 0001. By chance, this is
still a really old date, so no one noticed. However, with the port of
Emacs's fde99c729c (Port recent org-clock fix to POSIX time_t,
2018-03-28), test-org-clock/clocktable/ranges would fail if the system
supports the oldest date tried, "<-67715-09-22 Tue 17:51>".
But this "encode-time -> format-time-string -> org-parse-time-string"
dance is mostly unnecessary. All the current org-clock-special-range
callers except for org-clocktable-steps (1) explicitly check if the
starting time is nil, (2) don't use the starting time, or (3) pass it
directly to org-clock-sum, which handles nil values. And
org-clock-sum executes the same codepath when nil is passed instead of
"really old date".
Update org-clocktable-steps to use 2003 (the year Org was created) as
the starting point when org-clock-special-range returns nil for the
starting time. This is more efficient because we don't needlessly
calculate steps over a large chunk of time that almost certainly
doesn't have any clocked time. And it won't run into the portability
issues like Emacs's fde99c729c. 2003 _should_ be an appropriate
effective starting date, since we don't expect clocked time before the
existence of Org. If this turns out to be an issue (e.g., someone
converted reports from a pre-Org and still makes clock tables that
include those times), we can make this value configurable.
* lisp/org-clock.el (org-clock-resolve-clock): Store heading location
as marker instead of raw position to ensure that org-clock-in is
called in the correct buffer.
This fixes a regression introduced by 503ede74b (org-clock: Fix
resolving clocks, 2018-12-06).
* lisp/org.el (org-get-tags): Now org-get-tags returns tags list with
tags from #+filetags in the beginning.
* testing/lisp/test-org.el (test-org/get-tags): Add test.
Fixes regression caused by commit
<5e27b2fd32>.
Bug reported in
<https://lists.gnu.org/r/emacs-orgmode/2019-01/msg00052.html>.
* lisp/org.el (org-align-tags): Move point to the beginning of the
visible buffer first before attempting to look for Org headline
tags.
* testing/lisp/test-org.el (test-org/tag-align): Add test.
Fixes a regression in
1615261cdc.
Bug reported in
<https://lists.gnu.org/r/emacs-orgmode/2019-01/msg00051.html>.
* lisp/org-colview.el (org-columns--displayed-value): Add new optional
argument.
(org-columns--collect-values): Use new argument when called from
Agenda mode.
Reported-by: Allen Li <darkfeline@felesatra.moe>
<http://lists.gnu.org/r/emacs-orgmode/2018-12/msg00040.html>
* lisp/org-protocol.el (org-protocol-capture): Do not corrupt
`org-stored-links' when there is no URL to capture.
(org-protocol-do-capture): Remove function.
Reported-by: Allen Li <darkfeline@felesatra.moe>
<http://lists.gnu.org/r/emacs-orgmode/2018-12/msg00131.html>
* lisp/org-capture.el (org-capture-finalize): Do not save the target
capture file if :no-save keyword is non-nil in the capture template.
* doc/org-manual.org (Template elements),
lisp/org-capture.el (org-capture-templates): Document :no-save.
Ref: https://lists.gnu.org/r/emacs-orgmode/2018-09/msg00325.html
Paul Eggert
Kyle Meyer
Marco Wahl
Nicolas Goaziou
Kaushal Modi
Van L
Bastien
Renato Ferreira
Allen Li
Jens Lechtenboerger