* org.el (org-todo): Use marker `change-plist' position property to
permit triggering through "org-depend.el" with parent heading
statistics.
When TODO statistics in parent headings were updated it shifted the point
defining the :position property in `change-plist' causing headlines
with TRIGGER properties to fail to evaluate.
TINYCHANGE
* lisp/ob-exp.el (org-babel-exp--at-source): Source blocks generated
by evaluating other source blocks may not be found in the original
buffer. In that case, do not move point.
Reported-by: "Berry, Charles" <ccberry@ucsd.edu>
<http://lists.gnu.org/r/emacs-orgmode/2017-12/msg00538.html>
* lisp/ob-core.el (org-babel-expand-noweb-references): Get :noweb-ref
from definition instead of point of call.
* testing/lisp/test-ob.el (test-ob/noweb-expansion-1):
(test-ob/noweb-expansion-2): Merge tests. Add test.
* lisp/ob-scheme.el: Require geiser-impl so that
geiser-default-implementation is defined. Otherwise, unless Geiser
has previously been run (e.g. with `run-geiser'), you'll see an
error that geiser-default-implementation is undefined.
* lisp/ob-python.el (orb-babel-python-evaluate-session): When :session
:results output, send multiline code blocks to tmpfile and execute in
Python with exec().
(org-babel-python--exec-tmpfile): New variable.
* testing/lisp/test-ob-python.el (test-ob-python/session-multiline):
Test for :session with multiple lines and indentation.
* lisp/ob-haskell.el (haskell-prompt-regexp): Define defvar
`haskell-prompt-regexp`. This variable will override the variable set by
inf-haskell, when `org-babel-execute:haskell` is called.
(org-babel-execute:haskell): Make sure that
`comint-prompt-regexp` is set appropriately to enable correct parsing
of "λ"-prompts. Set `comint-preoutput-filter-functions` appropriately to
enable correct parsing of coloured ghci prompts.
The problem was that code sent back from the inf-haskell buffer to
org-babel wasn't parsed correctly in `org-babel-comint-with-output`.
This occured when the user uses the commonly used "λ"-prompt.
TINYCHANGE
* lisp/ob-gnuplot.el (org-babel-expand-body:gnuplot) if the Org buffer
is associated with a file, direct gnuplot to change to the directory
for that file.
* lisp/org.el (org-return): Split the table before first column or
after last one.
* lisp/org-table.el (org-table-next-row): Remove code handling split.
* testing/lisp/test-org.el (test-org/return): Add test.
* lisp/org-gnus.el: (org-gnus-store-link) Add a save-excursion to
protect the position of point. Otherwise, when calling org-capture
in an article with a virtual group, the point is moved and the
region captured is incorrect.
* lisp/org-capture.el (org-capture--prompt-history-table):
(org-capture--prompt-history): New variables.
(org-capture-fill-template): Handle history list per prompt.
Reported-by: Karl Eichwalder <ke@gnu.franken.de>
<http://lists.gnu.org/r/emacs-orgmode/2017-11/msg00251.html>
* lisp/org-archive.el: (org-archive-subtree) Do not save buffer after
each archive. Saving the archive buffer after archiving each
subtree results in substantial slowdown and many writes to disk when
archiving an active region. This brings the behavior of
org-archive-subtree into line with org-refile, which does not save
the target buffer after refiling.
* lisp/ox-publish.el (org-publish-file-relative-name): New function.
* lisp/ox-html.el (org-html-link): Use new function.
* testing/lisp/test-ox-publish.el (test-org-publish/file-relative-name):
New test.
* lisp/ox.el (org-export-numbered-headline-p): Use a less opinionated
definition of inheritance.
* testing/lisp/test-ox.el (test-org-export/numbered-headline-p):
Update test.
* doc/org.texi (Export settings): Insist on UNNUMBERED inheritance.
It is now possible to number a headline within an unnumbered tree.
* lisp/org-table.el (org-table-recalculate): Fix typo in existing
`remove-text-properties' call instead of adding a new one.
Also, cleaning property before processing formulas is more robust,
e.g., if last process raised an error.
* lisp/org-table.el (org-table-recalculate): Clean `:org-untouchable'
property once we no longer need it.
* testing/lisp/test-org-table.el (test-org-table/formula-priority):
New test.
Reported-by: Ruy Exel <ruyexel@gmail.com>
<http://lists.gnu.org/archive/html/emacs-orgmode/2017-11/msg00159.html>
* lisp/org-clock.el (org-clock--oldest-date): Fix an issue when
compiling on macOS 10.6 with a western time zone (a continuation of
Bug#27736).
Further workaround for faulty localtime() under macOS 10.6
98288ecaffbc0947ec3193088172ea85ad5e42a5
Charles A. Roelli
Wed Nov 8 20:51:48 2017 +0100
* lisp/ox-ascii.el (org-ascii-item): Use better heuristics to
determine if contents should follow bullet or start a new line.
Contents follow bullet when the first contributing line is
a paragraph. As a consequence, the following snippet
-
#+html: ...
This is a paragraph.
is exported as
- This is a paragraph.
instead of
-
This is a paragraph.
previously.
* lisp/ox-ascii.el (org-ascii-quote-margin): Update docstring.
(org-ascii--current-text-width): Properly handle width of text within
nested plain-list items.
(org-ascii-item): Always start a new line for description contents.
Indent them by `org-ascii-quote-margin' characters.
* lisp/org-element.el (org-element--list-struct): Correctly handle end
of sub-lists and end of last item.
* testing/lisp/test-org-element.el (test-org-element/item-parser):
(test-org-element/plain-list-parser): Add tests.
* lisp/org-capture.el (org-capture-mark-kill-region): Set end marker
insertion type so that it advances when inserting text at the end of
the capture buffer.
* testing/lisp/test-org-capture.el (test-org-capture/insert-at-end-abort):
Add test case.
Inserting text at the end of the capture buffer does not advance the
end kill marker. This causes the narrowing region and the kill
markers to get out of sync, which prevents 'org-capture-finalize' from
correctly cleaning the target buffer when aborting the capture.
Setting the kill end marker insertion type fixes this.
TINYCHANGE
* lisp/ob-tangle.el (org-babel-tangle-single-block): Prevent double unescaping
of source block by removing unnecessary call to `org-unescape-code-in-string'.
* testing/lisp/test-ob-tangle.el (ob-tangle/nested-block) New function.
* lisp/org.el (org-insert-todo-heading): Insert TODO keyword relative
to the original headline, not to the headline above.
* testing/lisp/test-org.el (test-org/insert-todo-heading-respect-content):
Add test.
* lisp/org-clock (org-clocktable-steps): Fix weekly-step clock report
ignoring hours between start of period and start of week, when start
of period is on a day of week numerically less than start of week.
Clock report now always inserts a "week" starting at the start of
the period, if necessary; all following weeks start on the start of
the week as expected.
TINYCHANGE
* lisp/org-capture.el: (org-capture-fill-template) Fix interactive
timestamp entry to match the behavior of org-time-stamp. This fixes
a bug related to the %^t and %^u template escapes which resulted in
incorrect timestamps (<2017-10-06 Fri-12:00>). The difference
between %^t and %^T now corresponds to the difference between
org-time-stamp called without and with a prefix argument.
* lisp/org.el (org-backward-paragraph): Do not error when called from
a block opening line.
* testing/lisp/test-org.el (test-org/backward-paragraph): Add tests.
Reported-by: Omar Antolin <omar.antolin@gmail.com>
<http://lists.gnu.org/archive/html/emacs-orgmode/2017-10/msg00051.html>
* lisp/org-clock.el (org-clock--oldest-date): Only execute
'decode-time' on times later than year -2**31 under macOS 10.6.
See Bug#27706.
Workaround for faulty localtime() under macOS 10.6
bd49b6f1b39cffeaf6098bc7b0182552683b1c07
Charles A. Roelli
Sun Oct 1 19:07:22 2017 +0200
* lisp/org-colview.el (org-columns-remove-overlays): Fix
"wrong-type-argument markerp nil".
When calling `org-agenda-columns', `org-columns-top-level-marker' is
not set. Handle this situation.
Reported-by: Benjamin Beckwith <bnbeckwith@gmail.com>
<http://lists.gnu.org/archive/html/emacs-orgmode/2017-09/msg00566.html>
* lisp/ox-publish.el (org-publish-get-base-files): Ensure MATCH regexp
is always a string since `directory-files-recursively' requires it.
* testing/lisp/test-ox-publish.el (test-org-publish/get-project-from-filename):
Add test.
Suggested-by: Jens Lechtenboerger <lechten@wi.uni-muenster.de>
<http://lists.gnu.org/archive/html/emacs-orgmode/2017-09/msg00528.html>
* lisp/ox-html.el (org-html-src-block): Unless klipse is used, export
source code blocks as content of `pre' elements, and not as content of
`code' children of `pre' elements. Restores the previous way of
exporting source code blocks, and fixes the display of language
indicators while hovering over the blocks in the exported HTML file,
when klipse is not used.
(org-html-keep-old-src): Remove it. Not needed because it is now the
default, unless klipse is used.
Emacs's bc511a64f6d made tree-wide changes to prefer https to http.
For Org, this resulted in the header of every lisp/*.el file being
changed. Make the same changes in lisp/*.el files that were added
since Org version 9.0.10 (the last version that was synced with
Emacs).
* lisp/ob-tangle.el (org-babel-tangle-publish): Port to new copy-file
behavior.
Adjust ob-tangle to new copy-file behavior
74b8615fcceba7b92c4938e1bcc92015f10ae899
Paul Eggert
Sun Sep 10 22:31:23 2017 -0700
* lisp/ob-scheme.el (org-babel-scheme-execute-with-geiser): Return
"#<void>" or "#<unspecified>" when the REPL says so.
* testing/lisp/test-ob-scheme.el (test-ob-scheme/unspecified): New
test.
(test-ob-scheme/prologue): Fix tests.
Reported-by: Lawrence Bottorff <borgauf@gmail.com>
<http://lists.gnu.org/archive/html/emacs-orgmode/2017-09/msg00346.html>
* lisp/org-capture.el: (org-capture-place-entry) Ensure that level is
set to 1 (i.e., the top level) if there is no headline target
defined. Otherwise, captured items are improperly nested by context.
* lisp/org.el (org-insert-link): Simplify so that description is only
prompted for once, if auto-desc is not set, and takes as its default
value, in order, default-description, the return value of
org-make-link-description-function (if the variable is non-nil), and
the current desc. Update the docstring to reflect that
default-description takes precedence over
org-make-link-description-function.
(org-make-link-description-function): Remove from docstring the
statement that if the variable is nil, then the link will be used as
the default description. This is undesirable, and was not in any case
implemented.
* lisp/ob-scheme.el (org-babel-scheme-execute-with-geiser): Use
Geiser's explicit 'retort' object, instead of catching and parsing
an expected Emacs message (which for me is in any case overwritten
by a following "Mark set" message); this also means we don't need
the with-output-to-string approach for getting 'output' results.
Also 'let' Geiser variables so as to avoid popping up the REPL
and *Geiser dbg* buffers.
* lisp/org.el (customize-package-emacs-version-alist): Add entries for
all existing ":package-version" values, and remove entries for
versions that aren't used as ":package-version" values.
Note that this alist has a limited effect because we specify
":version" in almost all of our defcustoms. Two functions make use of
this alist: describe-variable-custom-version-info and
customize-changed-options. The former only looks at the alist if
:version isn't specified, and the latter only looks at :version if an
entry for :package-version isn't found in the alist.
We should consider removing the defcustom :version specifications so
that the package-version->version mapping only needs to be maintained
in one place.
* lisp/org-duration.el (org-duration-to-minutes): Accept the empty
string.
* testing/lisp/test-org-duration.el (test-org-duration/to-minutes):
Add test.
For backward compatibility with `org-duration-string-to-minutes'
accept the empty string as a duration.
* lisp/ob-sqlite.el (org-babel-sqlite-expand-vars): Stop interfering
with the 'orgtbl-to-csv' default formatting.
* testing/lisp/test-ob-sqlite.el: New file.
The problem here was that ob-sqlite was providing a format to the
`orgtbl-to-csv' that actually overwrote the default format in that
function, and in the case where there were commas in the content of
the table, broke the table as argument mechanism.
TINYCHANGE