* org-capture.el (org-capture-fill-template): Use `org-set-property'
directly.
* org.el (org-set-property): Split property and values reading.
(org-read-property-name, org-read-property-value)
(org-set-property-function): New functions.
(org-property-set-functions-alist): New variable.
The goal of this patch is to introduce a special variable
`org-property-set-functions-alist'. This variable allows to read
properties values in a more intelligent way from `org-set-property' or
from `org-capture'.
For that, it simplifies the `org-set-property' code and remove
duplication between `org-capture' and `org-set-property'.
Signed-off-by: Julien Danjou <julien@danjou.info>
* lisp/org.el (org-make-target-link-regexp): regexp-quote target
before replacing whitespace.
Previously a radio link <<<...>>> would match all three-letter words
in the buffer. The manual indicates the radio links are meant to
match literally (modulo whitespace differences), so we should
regexp-quote all the targets to avoid over-eager matching.
* lisp/org-latex.el (org-export-latex-first-lines): Anchor outline
regexp during LaTeX tree export
Jrg Hagmann writes:
> - If you export the (new) minimal example below to latex (C-cC-e l), it works.
> - If you only export a tree (Subtree in the example; C-cC-e 1 l), the first table ends at the horizontal line and everything between it and the next node (Subsubtree) is eliminated. The second (identical) table is exported correctly.
> - If you remove the asterisk(s) in the first table, it works.
>
> This problem crept in in the last days or weeks before 7.4.
>
> It may not be a problem for most of you, but I happen to have a number of files where columns are automatically displayed as tables preceding the first subnode (#+BEGIN: columnview ...). An alternative would be to display %ITEM in column-view without the asterisks.
>
> Emacs 23.2.1 on OS X 10.6.5
> Org-mode version 7.4 (release_7.4.24.g48b11.dirty)
>
> Thanks, Jrg
>
> -------New minimal example------------
> * Subtree
>
>
> | One | Two | Three |
> |--------+------+-------|
> | * Test | text | text |
> | ** One | text | text |
>
>
> Some text
>
> ** Subsubtree
>
> | One | Two | Three |
> |--------+------+-------|
> | * Test | text | text |
> | ** One | text | text |
* lisp/ob-python.el (org-babel-python-initiate-session-by-key): Make
sure that py-which-bufname is initialized, as otherwise it will be
overwritten the first time a Python buffer is created.
* lisp/org.el: (org-entry-properties) Stop scanning for timestamps if
a specific timestamp property (e.g., DEADLINE, SCHEDULED, etc.) is
requested and a match is found. Also, if a specific timestamp property
is requested, do not push non-relevant timestamps onto property list.
This change only effects org-entry-properties when a specific
timestamp is requested with the special flag, as in:
(org-entry-properties nil 'special "SCHEDULED")
Previously, even if only the SCHEDULED timestamp was requested,
org-entry-properties would parse all the timestamps in an entry. This
extra parsing could slow down the construction of agenda views,
especially with entries that contained a large number of log
items (CLOCK, state changes, etc.). The function org-entry-get,
however, is only interested in the first occurrence of the item. When
looking for a specific type of timestamp, org-entry-properties now
stops searching for timestamps after the match is found, unless the
property is "CLOCK".
Here are the relevant ELP results:
Before:
org-entry-get 296 0.4724579999 0.0015961418
org-entry-properties 31 0.3438769999 0.0110928064
After:
org-entry-get 296 0.1447729999 0.0004890979
org-entry-properties 31 0.015765 0.0005085483
* lisp/org-agenda.el: (org-agenda-get-scheduled) Don't call
org-is-habit-p until after checking for for
org-agenda-skip-scheduled-if-done.
Org-agenda-get-scheduled was calling org-is-habit-p on every scheduled
item (including DONE items when org-agenda-skip-scheduled-if-done was
set to t). Tweaking the timing of the test shaves some time off of
agenda construction when org-habit is loaded and
org-agenda-skip-scheduled-if-done is t.
Before: org-is-habit-p 478 0.2434439999 0.0005092970
After: org-is-habit-p 81 0.057944 0.0007153580
* org.el: remove spurious linebreak introduced by earlier patch
* ob.el, ob-ref.el: remove double fix of the same problem
Achim Gratz <Stromeko@Stromeko.DE> wrote:
> this patch had already been partially applied by Carsten and Eric (in
> slightly a different way than I suggested). The changes to ob.el and
> ob-ref.el (the require statements) are therefore superfluous and should
> probably be backed out. There was also a superfluous whitespace change
> in org.el (a closing paren that was broken onto the next line). Patch
> to this effect is attached. You've already cleaned up org-agenda.el and
> the conditions in org-macs...
* lisp/org-html.el (org-export-html-mathjax-template): displaymath
environment and MathJax
Greetings All.
The following patch makes MathJax consider \begin{displaymath} and
\end{displaymath} as math environmetn boundaries. For someone who, like
me, keeps "The not so short introduction to LaTeX2e" alway around, the
displaymath environment is the default way to introduce a block of math.
In fact '\[' and '\]' are also mentioned there but the environment is
used in every single example so the patch minimizes the surprise.
* lisp/org-faces.el (org-agenda-current-time): New face.
* lisp/org-agenda.el (org-agenda-show-current-time-in-grid):
(org-agenda-current-time-string): New options.
(org-agenda-add-time-grid-maybe): Add current time to time grid.
suvayu ali <fatkasuvayu+linux@gmail.com> writes:
> I actually tried to set the text properties for the string instead,
> but looks like org-agenda is ignoring that.
>
> (defun jd:org-current-time ()
> "Return current-time if date is today."
> (when (equal date (calendar-current-date))
> (propertize (format-time-string "%H:%M Current time") 'font-lock-face
> '(:weight bold :foreground "DodgerBlue4" :background "snow"))))
To accomplish this you'd have to apply the following patch and use 'face
property rather than font-lock-face.
Why can't a sexp choose its 'face after all?
--8<---------------cut here---------------start------------->8---
--8<---------------cut here---------------end--------------->8---
* org-footnote.el (org-footnote-create-definition): Place Footnotes
section before message-signature-separator also in modes derived
from message-mode.
* lisp/org-list.el (org-list-top-point-with-indent,
org-list-bottom-point-with-indent): Pay also attention to
'original-indentation property of text, as blocks are put to column
0 upon exporting.
* lisp/ob.el (org-babel-remove-temporary-directory): Handle exception
with message informing of failure to remove directory.
Thanks to Antti Kaihola for the bug report:
http://thread.gmane.org/gmane.emacs.orgmode/34394
From: Antti Kaihola <akaihola <at> gmail.com>
Subject: Can't close Emacs+org-mode if /tmp and /home on different partitions
Newsgroups: gmane.emacs.orgmode
Date: 2010-12-02 08:33:28 GMT (6 days, 1 hour and 22 minutes ago)
I have /tmp on my root partition and a separate partition for /home.
When trying to close an Emacs session which is using org-mode, I get
this error:
move-file-to-trash: Non-regular file: Is a directory, /tmp/babel-XXXXXXX
(where XXXXXXX are random characters).
I tracked down the problem to org-babel-remove-temporary-directory
which ob.el adds to kill-emacs-hook. It tries to remove the temporary
directory using delete-directory, which in turn tries to move the
directory (by renaming) into trash, which is in my home directory.
I added this to my ~/.emacs.d/init.el:
(custom-set-variables '(temporary-file-directory "/home/akaihola/tmp/"))
and closing Emacs works correctly again. However, since my init.el is
part of emacs-starter-kit which I update frequently, I'd prefer not to
modify that file. Unfortunately the customization hook
emacs-starter-kit provides (~/.emacs.d/custom.el) is loaded too late
to affect the temporary directory.
I'm running emacs-snapshot 1:20090909-1 in Ubuntu 10.10. Looks like
this is really an Emacs bug and is already fixed:
http://groups.google.com/group/gnu.emacs.bug/browse_thread/thread/0446b8684a8ef504
* lisp/ob-clojure.el (org-babel-header-arg-names:clojure): Adding
`package' to the list of Clojure header arguments which will be read
from heading properties.
* lisp/org-inlinetask.el (org-inlinetask-export-templates): added
Sébastien Vauban's suggestion for LaTeX export in docstring. This is
not default as it requires an additional LaTeX package: "todonotes".
* org-inlinetask.el (org-inlinetask-export-templates): new variable
* org-inlinetask.el (org-inlinetask-export-handler): make use of
templates to export inline tasks
* org-inlinetask.el (org-inlinetask-outline-regexp): new function
* org-inlinetask.el (org-inlinetask-goto-beginning): new function
* org-inlinetask.el (org-inlinetask-goto-end): new function
* org.el (org-mark-subtree): new command
* org.el (org-speed-commands-default, org-mode-map): make use of new command
This reverts commit 383802d063.
The commit had org-mode process the headlines from bottom to top, which
meant that any changes to the visibility of lower headlines were
overridden/modified by changes higher up the tree. Reverting the commit
causes VISIBILITY to work correctly.
* lisp/org.el: (org-make-heading-search-string) Optionally limit
number of lines stored in file link search strings.
(org-context-in-file-links) Add option to set to integer specifying
number of lines.
B S will cause tasks to be rescheduled a random number of days into the
future, with 7 as the default. This is useful if you've got a ton of
tasks scheduled for today, you realize you'll never deal with them all,
and you just want them to be distributed across the next N days.
* org-agenda.el (org-agenda-custom-commands-local-options):
Allow org-agenda-span to be a symbol.
(org-agenda-ndays): Make obsolete.
(org-agenda-span): New variable superseding org-agenda-ndays.
(org-agenda-menu): Use org-agenda-current-span.
(org-agenda-current-span): New local variable storing current
span.
(org-agenda-list): Take a span instead of ndays as argument.
This function is now responsible for computing the ndays based
on span.
(org-agenda-ndays-to-span): Return span only if number of days
really matches.
(org-agenda-span-to-ndays): New function.
(org-agenda-manipulate-query): Use org-agenda-compute-starting-span.
(org-agenda-goto-today): Use org-agenda-compute-starting-span.
(org-agenda-later): Do not give compute a new span, use the
current one.
(org-agenda-day-view, org-agenda-week-view)
(org-agenda-month-view, org-agenda-year-view): Stop touching
org-agenda-ndays.
(org-agenda-change-time-span): Only compute starting-span.
(org-agenda-compute-starting-span): New function derived from
the old org-agenda-compute-time-span.
(org-agenda-set-mode-name): Compute mode based on
org-agenda-current-span.
(org-agenda-span-name): New function.
* org-mouse.el: Replace Replace org-agenda-ndays by
org-agenda-current-span.
* org.texi, orgguide.texi: Replace org-agenda-ndays by
org-agenda-span. Add a paragraph about org-agenda-span and say that
org-agenda-ndays is now deprecated.
This patch is pretty huge, so I'll give a bit of context about it.
I'm weird, but I used org-agenda-ndays set to 14. Unfortunately, this
settings was interpreted as a month view. Pressing 'f' key to see later,
would show me the next month, which was not at all what I wanted.
On the same idea, day view or week view would change my org-agenda-ndays
settings, which I think is not a good idea. Changing user setting is *bad*.
:-)
So I rewrote the things this way:
- Rename org-agenda-span to org-agenda-current-span
Which has the same meaning has before, except it can be numeric.
- Rename org-agenda-ndays to org-agenda-span
I think the name is better choosen. You can set it to a symbol instead of
only a numeric value. That means you can set it to 'month and it will show
you the number of days of the current month in your agenda. Better than
30. But you can still set it to 30, or 31, or whatever you want.
- Do not change org-agenda-span. Never.
- Use org-agenda-current-span for navigation.
That means if you press 'f', it will shows you really the next
org-agenda-current-span ndays, and not something based on "I think you
want a weekly view".
* lisp/org-agenda.el: (org-format-agenda-item) The value of
org-category is not converted to a string unless it is defined.
This fixes commit 3061c7083d, which
resulted in org-format-agenda-item always returning the symbol-name for
org-category even if it was not defined. I.e., in some instances,
org-format-agenda-item returned the string "nil", thus bypassing the
buffer-file-name method of deriving the category.
* lisp/ob-python.el (org-babel-execute:python): Pass the new "prefix"
header argument through to external evaluation.
(org-babel-python-evaluate): Pass the new "prefix" header argument
through to external evaluation.
(org-babel-python-evaluate-external-process): When specified prepend
"prefix" to the file used in external evaluation.
* lisp/ob-sql.el (org-babel-expand-body:sql): Expand the body of a sql
code block.
(org-babel-execute:sql): Use sql specific body expansion function.
(org-babel-sql-expand-vars): Insert variables into a sql code block.
* lisp/ob-sqlite.el (org-babel-execute:sqlite): Remove unused variable
declaration.
* lisp/org-clock.el (org-day-of-week): New function.
(org-quarter-to-date): New function.
(org-clock-special-range): Implement quarters.
Patch by Erwin Vrolijk
* org-agenda.el (org-format-agenda-item): Convert category to a string
if it is a symbol. This fixes the following call to
org-agenda-get-category-icon which fails if category is not a string.
Signed-off-by: Julien Danjou <julien@danjou.info>
* lisp/ob-clojure.el: Updated requirements documentation to mention
the minimum version of Clojure.
(org-babel-expand-body:clojure): Fully qualified function name.
* lisp/org-latex.el (org-export-latex-lists): do not add an
unnecessary newline character after a list.
* lisp/org-list.el (org-list-bottom-point-with-indent): ensure bottom
point is just after a non blank line.
* lisp/ob-python.el (org-babel-execute:python): Use a :return header
argument for external evaluation in which the code block body need
be wrapped in a function
Thanks to Darlan Cavalcante for proposing this feature.
Julien Danjou
Eric Schulte
Carsten Dominik
Lawrence Mitchell
John Wiegley
Bastien Guerry
Konrad Hinsen
Matt Lundin
Achim Gratz
Łukasz Stelmach
Tassilo Horn
Nicolas Goaziou
Dan Davison
Robert Pluim
Sébastien Vauban
Allen S. Rout