All export commands now push the result to the kill ring by default.
This is subject to the variable `org-export-push-to-kill-ring'.
Also, this commit adds a new variable
`org-export-show-temporary-export-buffer' which can be used to turn
off the display of the temporary buffer containing the exported text.
Since this stuff is now automatically pushed onto the kill ring, some
people might prefer not to see this buffer.
Mikael Fornius writes:
> This is my suggestion of an implementation of min/mean/max computation
> in columnview summaries. If you like it feel free to use it.
>
> New operators: {min}, {max} and {mean} possibly prefixed with : for use
> with timevalues.
>
> Example from my running exercise diary:
>
> #+COLUMNS: %DISTANCE{+;%.1f} %HEARTRATE{mean;%.1f} %SPEED{:min} %CALORIES{+}
>
> Gives a colview with summaries:
>
> total distance, mean heartrate, fastest speed (min/km) and total
> calories.
>
> I have tested it on emacs-23 and it works well for me now, also with the
> interactive colview functions.
>
> But you never know really. ;-) Anyway, there should not be any emacs-23
> specific elisp code added afik.
>
> (Because I do not use xemacs I have not tested it with xemacs but the
> small changes I made should be compitable to both xemacs and emacs. I
> would appreciate if someone on this list who uses xemacs will give it a
> try for me. Thanks!)
>
> (This fix also opens up for using user defined lisp functions to
> calculate colview summaries, but I am not sure if that is something
> useful. Like this:
>
> (defun std (&rest values)
> "Compute standard deviation."
> ...)
>
> #+COLUMNS: %DATA{eval:std}
>
> If someone finds this attractive it would now be easy to implement as well.)
>
Setting the new option `org-hierarchical-todo-statistics' to nil will
make TODO statistics to be computed recursively. This means, not only
the direct children of a node contribute to its TODO statistics, but
the entire subtree.
You can also set the COOKIE_DATA property and add the word "recursive"
there to get recursive statistics for a specific tree.
Publishing a project with prefix argument to
force publishing was broken. Actually, the entire interactive side of
this function was implemented badly. Fixed now.
Hsiu-Khuern Tang writes:
> Hi Carsten,
>
> You recently changed org-edit-src-code to use a separate buffer
> instead of an indirect buffer. One side effect of this is that I
> can no longer edit several code examples at the same time:
> opening the second buffer will silently discard any changes made
> in the first. I would prefer this behavior: when opening the
> second edit source buffer, write any changes in the first buffer
> to the originating Org buffer (but don't save it, of course).
>
> Another approach is to use different buffer names.
The better approach is clearly to allow several buffers, now
implemented with this commit.
Harri Kiiskinen writes:
> There is an elusive bug in the ICal export functions. To
> demonstrate it, open this file, allow for the use of the local
> variable, and run 'org-export-icalendar-this-file'. What I get is
> the message "Wrong type argument: stringp, nil"
>
> -----------------------------
> #+FILETAGS: test
> * TODO header
>
> # Local Variables:
> # org-use-tag-inheritance: "^[t]"
> # End:
> -----------------------------
>
> Now, the funny thing is, that if you change the filetag from
> "test" to "est", close the file, and open it again, the file will
> be exported fine! It seems, that if the filetag does match the
> regular expression in org-use-tag-inheritance, then the result is
> error.
This bug was due to a missing save-match-data around a call to
org-get-tags-at. However, when studying this bug it turned out
that explicitly checking for the archive tag is not needed,
because we already have a call to `org-agenda-skip' in the same
loop.
Peter Frings writes:
> Good day, org-ers
>
> I have seen a *small* problem in the Agenda's log mode.
>
> Org-mode version 6.26d
> GNU Emacs 22.3.1 (i386-apple-darwin9.5.0, Carbon Version 1.6.0) of 2008-11-01
>
>
> I have this in time-sheets.org:
>
>
> * Project AI
> ** Enhancement PT <2009-04-30 Thu 10:00-11:00 +1w>
> CLOCK: [2009-05-14 Thu 09:55]--[2009-05-14 Thu 10:45] => 0:50
>
>
> But, the Agenda shows this:
>
> Day-agenda (W20):
> Thursday 14 May 2009
> 9:00...... ----------------
> planning: 9:12- 9:55 Clocked: (0:43) BUSY Notifications
> time-sheet: 10:00-11:00 Clocked: (0:50) Enhancement PT
> time-sheet: 10:00-11:00 Enhancement PT
> 10:00...... ----------------
> 11:00...... ----------------
>
He is right. Progress lines should never take their
time from the headline, always only from the triggering line.
Benjamin Andresen writes:
> I was curious as to why a value such log-mode wasn't customizable
> to be run on start-up as follow-mode and clock-report-mode are.
>
> If Carsten thinks this is a good idea, I've attached a patch that
> may shave of 5 seconds of him adding it. It applies cleanly to
> current git HEAD.
This commit applies Beanjamin's patch
There is now a new export function, `org-export-as-org', which
basically creates a copy of the Org file with things like archived
trees, commented trees, and trees deselected by export tags,
stripped.
This commit adds:
{{{date(FORMAT)}}} current date/time, formatted with
`format-time-string'
{{{modification-time(FORMAT)}}} date/time of last modification of
file, formatted with `format-time-string'
{{{input-file}}} the file name of the source Org file.
entries.
Guy Wiener writes:
> Hello everyone,
>
> I use orgmode to write down TODO tasks with dependencies (using
> org-enforce-todo-dependencies). I want to export the tasks to an
> iCalendar file, but *without* the blocked tasks (i.e, tasks that have
> unfinished dependencies). The agenda view hides these tasks if you set
> org-agenda-dim-blocked-tasks. It there a way to configure the
> iCalendar export to ignore these tasks too, like in the agenda view?
This commit implements this, when the value of
`org-icalendar-include-todo' is the symbol `unblocked'.
The commit implements a general mechanism for providing special
(e.g. completion) support for specific link types when entering links
with `C-c C-l'. After calling `C-c C-l', you may now press RET after
inserting a link prefix, and Org will look for a function
`org-PREFIX-complete-link'. Such functions may be defined for any
link types, including link abbreviations. Currently, Org has
`org-file-complete-link' for file name completion, and
`org-bbdb-complete-link' for completion of record names from BBDB.
Peter Westlake writes:
> On Fri, 8 May 2009 06:24:54 +0200, "Carsten Dominik"
> <carsten.dominik@gmail.com> said:
>
> I anyone could make an example that allos me to reproduce this
> problem, then I might be able to fix it.
>
> It's trivial:
>
> * top
> ** sub 1
> Set org-cycle-include-plain-lists.
> Type TAB on this line, and the whole of top gets folded.
>
> ** sub 2
> This entry is not needed to reproduce the bug. It just
> makes it more visible.
There was indeed a bug, I think it is fixed now.
Shaun Johanson writes:
> Consider the following Org file:
>
> * Test
>
> See [[(foo)][FOOBIE]]
>
> #+BEGIN_EXAMPLE
> <foo>: blah blah (ref:foo)
> #+END_EXAMPLE
>
> Question 1)
> In Org mode the link displays as FOOBIE, in the exported HTML it
> displays as (foo). Is there any way to cause the link to use the
> description (FOOBIE) in HTML? If not would this be a useful option
> to add?
This was a bug, fixed now.
Editing source code examples and pictures is not done in a truly
separate buffer, not in an indirect buffer. Indirect buffers had
caused problems with fontification, for example.
Jason Riedy writes:
> I'm trying to change org-export-latex-image-default-option
> to "width=.7\\linewidth" in a file local variable. It's set
> correctly as a buffer local variable, and it's having no
> effect on the export. My guess is that the buffer-local
> property is stopping it as soon as org-export-as-latex runs
> set-buffer.
>
> I can smuggle the value in by adding an entry to org-export-plist-vars
> referring to org-export-latex-image-default-option and pulling the value
> from the plist, but that feels incorrect.
It is actually the correct way to do this, and I have
implemented this change.
Samuel Wales writes:
> I frequently export to ascii without wanting a file to be created,
> especially not in a useful directory, as the files are temporary.
>
> Is there a way to export ascii to just a buffer?
There is now, `C-c C-e A'.
This commit also implements commands
- org-export-as-ascii-to-buffer
- org-replace-region-by-ascii
- org-export-region-as-ascii
which are similar to what is available for HTML and LaTeX.
`C-c C-e A' used to be the key for publishing all projects.
This functionality has now been moved to `C-c C-e E'.
Scot Beker writes:
> I have org-remember working successfully (I can't believe that it took
> me so long to experiment with it. It has made a huge difference in my
> work).
>
> But if, after calling the extension with C-c r, I type a "wrong" tag
> selector (e.g. I mistype a right one), I get
> (1) the error message 'no template' (good)
> (2) a remember buffer without a bullet point, with the point at the
> very top of the file (sure, what do I expect?)
> (3) but, having realized my mistake, I can't get out of it gracefully.
> I can't abort (C-c C-k), and can't (C-c C-c).
> I get the error message "Args out of range 0, 1"
>
> Any idea whether I've set something up wrong or if org just has no
> mercy on those who can't manage to type one of the selected values?
> Otherwise, org-remember works well.
This situation is now avoided by forcing the selection of a valid
template.
Matt Lundin writes:
> When I select a region and invoke
> org-replace-region-by-latex, the region is removed, but no
> latex output is put in its place. In other words, the region
> is simply deleted.
>
> Strangely, if I select multiple headlines, they are
> converted to latex. But if I select only text underneath a
> headline, it is not replaced.
This is hopefully fixed now.
Alex Mizrahi writes:
> I'm trying to use column view to work with effort estimates, but it
> does not seem to work in XEmacs. When I press C-c C-x C-c columns
> show up in agenda view (I'm not sure if 100% correcly, but
> more-or-less so), but when I try editing anything, column view goes
> away and error appears:
>
> Wrong number of arguments: #<subr local-variable-p>, 1
Chris Gray had the idea to have arbitrary blocks turned in LaTeX
environments and HTML divs. These three new hooks allow
implementation has an add-on rather than a patch.
Thomas Morgan writes:
> I just tried exporting an Org file with LaTeX fragments to HTML
> on a computer that doesn't have dvipng. There were error messages
> in *Messages* ("Failed to create png file..."), but this wasn't
> obvious to me at first glance because those messages were replaced
> in the echo area by "Exporting... done" before I could see them.
>
> So I was wondering, is there a good way to make the user aware of
> those errors? Maybe by printing "Exporting... done (with errors)"?
There is now a better error message when either the latex or the
dvipng program does not exist.
Carsten Dominik writes:
> On Apr 26, 2009, at 12:08 AM, Samuel Wales wrote:
>
> > I noticed the following regarding
> > org-remember-backup-directory:
> >
> > 1) auto-saving does not save contents
> > - solution: put (auto-save-mode 1) before or after (not
> > sure which) the call to org-set-local in the
> > following lines in org-remember:
> >
> > # (expand-file-name
> > # (format-time-string "remember-%Y-%m-%d-%H-%M-%S")
> > # org-remember-backup-directory))
> > # (save-buffer)
> > # (org-set-local 'auto-save-visited-file-name t))
> > # (when (save-excursion
>
> Done.
>
>
> > 2) save-buffer is called before there are any contents
> > - this causes files without useful content to stick
> > around in the dir if the buffer is killed or
> > something.
> > - solution: don't save the buffer; let auto-saving do
> > that work. will this work?
>
> Usually, the template expansion will have filled in stuff like
> links and initial contents, so I would prefer to keep the
> save-buffer here and let the user deal with the extra files.
>
> > 3) if you have auto-save-file-name-transforms set,
> > auto-saving is done according to that variable instead
> > of saving in place.
> >
> > I am now turning that variable off, locally in the remember buffer.
> > Does that fix the problem?
>
> - Carsten
>
> > - solution: is this an emacs bug? i am not sure. the
> > auto-saving should be done in place, saving in
> > org-remember-backup-directory and not saving in the
> > place where the transforms variable tells it to save.
> > perhaps each user can kludge transforms or perhaps
> > emacs needs fixing? i have reached debugging ability
> > limit.
Tassilo Horn writes:
> I have this entry in an org-file.
>
> --8<---------------cut here---------------start------------->8---
> ** bung Grundlagen der Softwaretechnik (B 016) :teaching:
> <2009-04-22 Wed 08:30-10:00 +1w>
> <2009-04-29 Wed 10:15-11:45> (Fr Hannes bernehmen)
> --8<---------------cut here---------------end--------------->8---
>
> I expect it to be displayed every wednesday after 2009-04-22
> for 08:30-10:00 and on 2009-04-29 it should be displayed
> twice: one at the usual time and one for 10:15-11:45.
>
> But the agenda shows only the regular entry (08:30-10:00)
> and omitts the exception. If I reorder the timestamps, so
> that the exception comes first, only the exception will be
> shown on 2009-04-29. Every other wednesday shows the
> regular repeater date.
>
> What do I have to do to make that entry showing up twice on
> 29th April?
This commit adds a new option
`org-agenda-skip-additional-timestamps-same-entry'. When set
to t, an entry that has multiple active time stamps will get
multiple lines in the agenda.
This reverts commit 61dfa0c67f.
This wasn't such a great idea after all. There is one case where
this is really annoying - if you are in the middle of displaying
an agenda view with lots of filters active and you change what you
are clocking it jumps away and you lose the agenda view.
This one case makes hitting a dedicated function key that runs
org-clock-goto more desireable.
Matt Lundin writes:
> If it's not too much trouble, I was wondering if I could
> request the following properties to set export options for
> subtrees:
>
> EXPORT_AUTHOR
> EXPORT_DATE
>
> In addition to specifying an EXPORT_FILE and EXPORT_TITLE
> for a subtree, I often find myself wanting to change the
> date and author lines.
The variable `org-refile-target-verify-function' can be set to a
function that will be called to verify a refile target. The function
must return t if the target is valid.
Users can now define custom IDs for use in HTML export.
These IDs are stores as property CUSTOM_ID. When present, HTML will
prefer using these over automatic targets like "sec-N.M".
Scot Becker writes:
> Prompted by Chris Gray's request for org markup in Latex
> environment, I thought I'd submit a note (for his sake and
> others') about a few quirks of org-latex-export's handling
> of embedded Latex markup in org documents. I have been
> puzzling with these for a while but only discovered the
> problem triggers (and workarounds) this morning just before
> Chris' mail arrived. These are both about inline Latex
> commands:
>
> I use a few custom commands \mycommand{like this}, and
> occasionally have to invoke the odd bit of standard LaTeX
> markup, for example /when \textbf{embedding bold text}
> inside italics/. For the most part, these work fine, but
> I've discovered the following two 'gotchas' that happen when
> exporting to LaTeX.
>
> 1. Inline Latex commands get their final curly brace
> escaped with a slash (and therefore don't work) if they
> spill over into another line, i.e. if they contain one or
> more newlines. This is true also for standard LaTeX
> commands like \textbf{} and \emph{}.
>
> ----------------SAMPLE------------------------
> \mycommand{So, for example this
> wrapped setence gets a slash added just after the
> final period and before the curly brace.} Org is quite
> helpfully escaping the slash for LaTeX, apparently.
>
> \mycommand{no trouble if it's all on one line}
> ------------------END-------------------------
>
> The workaround of putting all such commands on one line is
> no hardship for me, since I use visual-line-mode in Emacs 23
> and keep my paragraphs as single logical lines. It might be
> harder for those accustomed to hard-wrapping their
> paragraphs.
>
>
> 2. If you have two inline Latex commands on the same
> logical line, org's latex export doesn't treat the text
> between them in its usual manner. Italics get processed,
> but not the latexification of quotes. ("this" --> ``this'')
> For example:
>
> ----------------SAMPLE------------------------
> I have a short custom command to tell Latex to invoke a
> Hebrew-language right-to-left environment when I want to refer to a
> Hebrew phrase like this: \heb{phrase here}. But then if I "quote
> something," and follow that by another \heb{phrase}, the inner
> quotation marks don't get processed. Oddly enough, this problem is
> only triggered when there is an inline Latex command both before and
> after the quoted material on the same logical line.
>
> Now if you put a footnote in between those two inline Latex commands,
> the output is really nutty:
>
> And \heb{phrase here} with a footnote[fn:: Footnote here.] I'm not
> sure what funky org commands get invoked, but again, only when
> bookended by an inline Latex command like \heb{phrase here}.
> ------------------END-------------------------
>
> The nutty output is a number in square brackets like
> this[1], with the following at the bottom of the document:
>
> \$\^{}{1}\$ Footnote here.
>
> This has a the opposite work-around: break the lines so
> those elements are not all on the same logical line. Put in
> a few newlines. Latex, of course doesn't care. Do take
> care not to start a newline with the org-footnote, like this
> [fn:: Org doesn't parse a footnote command which starts on
> its own line.]
>
> This is just "for what it's worth" to those who use org-mode
> as a front-end to writing for LaTeX.
These problems were caused by a regular expression for
matching latex macros with arguments, that did not allow any
newlines. Now we have a much better regexp, that even
allows for three levels of nested braces.
Move the point to the selected task when clocking in using the clock
history. I find I'm always going to the currently clocked task after
picking it off of the menu and this saves one small step.
This could be optionally controlled by a variable.
Sometimes refiling a task displays the next task heading after ... at the
end of a folded task. This keeps the next task the cursor is on starting
in column 1 which feels more natural.
If the heading field in the remember template entry is either `top' or
`bottom', it is now OK to file to a file that is not in org mode, and
the content of the remember buffer is inserted without forcing an
Org-style header.
Rares Vernica writes:
> I think the standard references do not work correctly in the
> "remote" function. Moreover, the "edit all formulas" (C-c ')
> window replaces the internal references with standard
> references. Even if I toggle the references back to internal
> ones, the references in the "remote" function do not get
> updated.
>
> Here is an example:
>
> #+TBLNAME: TableA
> | 101 |
> #+TBLFM: @1$1=remote(TableC,@1$1)
>
> #+TBLNAME: TableB
> | A1 |
> #+TBLFM: @1$1=remote(TableC,A1)
>
> #+TBLNAME: TableC
> | 101 |
>
> If I do C-c * in TableA, it works correctly. In TableB it
> doesn't. If I do C-c ' in TableA and then (with or without
> C-c C-r) C-c C-c and C-c *, then the contents of TableA will
> be equivalent to the ones of TableB and the reference will
> be broken.
Standard references like A1 are now allowed in call to
remote().
Rares Vernica writes:
> I think I found another bug related to remote
> references. When I insert/remove a row/column using the
> table commands, the remote references to other tables are
> also updated. I think org treats "remote" as a regular
> function and updates the references inside it.
>
> Here is an example:
>
> #+TBLNAME: TableA
> | 101 |
> #+TBLFM: @1$1=remote(TableB,@1$1)
>
> #+TBLNAME: TableB
> | 101 |
>
> If I go in the cell of TableA and do M-S-down arrow, I get
> the following:
>
> #+TBLNAME: TableA
> | |
> | 101 |
> #+TBLFM: @2$1=remote(TableB,@2$1)
> ^^^^
>
> As you can see the remote reference has been updated. I
> similar update happens when I remove a row or insert/remove
> a column.
This commit makes sure that references inside calls to
remote() are not touched.
sha1-string is not autoloaded in sha1.el as in the version distributed
with Emacs 22. Instead of relying on autoloads, the sha1 library is
now required by org-feed.el.
When refiling, you can now create new parent nodes on the fly. To do
this, set the variable `org-refile-allow-creating-parent-nodes' to
`confirm'. Then, at a refiling prompt, proceed with completion until
you have an existing heading, and then add "/new heading", i.e. a
slash followed by the new heading. That heading will be created as a
child of the existing heading, and the entry to be refiled will end up
under that new heading.
New wrapper span around keyword plus time stamp, with class
timestamp-wrapper.
.timestamp-wrapper {float: right;}
could be a nice entry in a CSS style file.
Adam Elliot writes:
> Automatically resuming the clock after an Emacs restart
> fails under the following cases:
>
> 1. If org-log-states-order-reversed set to t (default), and
> a state change line precedes the clock line to resume.
> Error message is "Cannot restart clock because task does
> not contain unfinished clock".
>
[...]
> 2. If org-log-states-order-reversed set to nil. Error
> message is the same. Reason: point is placed *after*
> last clock line and so fails looking-at test.
>
This commit fixes the problem, in a slightly different way
than Adam proposed. Instead of trying to fix the old way to
find the position of the clock, we now simple search the
entry if there is an unfinished clock and go there. Since
new clocks are added before older ones, this should be a
safe bet.
Time stamps in LaTeX export now also honor custom time stamp formats.
Furthermore, the new option `org-export-latex-timestamp-markup' can
specify special markup for time stamps.
Some of the standard export options are now defined in backend
specific files. This commit makes sure that building the options
property list will not cause an error because of unneeded (for the
backend) undefined variables.
Samuel Wales writes:
> A lower case version of a todo kw at the beginning of a
> headline, when in lower case, causes sort to ignore the
> word.
>
> Also, setting priority with shift down causes the cookie to
> be inserted in the wrong place.
Both problems are address in this commit.
Remember and refile processing does not require a task. This change
removes the unneeded default task.
This supports a workflow where new remember tasks and notes go into a
mostly empty file which just has #+FILETAGS: at the top and nothing
else.
This workflow has a minimal number of remember templates
- one for new tasks and (filed in tasks.org)
- one for new notes (filed in notes.org)
- one for phone calls (filed in phone.org)
New tasks are added as top-level tasks to the end of these files and
the #+FILETAGS: REFILE header causes each task to be easy to find.
All tasks in these files are refiled to a more appropriate org file at
a later time.
The following contributed packages are (partially) obsolete.
org-browser-url.el
org-annotation-helper.el
The functionality of both these packages is a subset of
org-protocol.el, which is now part of the Emacs core
and is recommended.
org-depend.el
A significant fraction of the org-depend functionality
dependence on siblings, children, and parents) is now
built-in into the Org core. Org-depend remains
in the distribution as a proof-of-concept fro complex
and remote dependencies.
org-interactive-query.el
I believe that much of what this package was build for
is now available with tag filtering.
These packages are now marked in org-modules as such.
The command org-reload did not only reload any loaded files, but all
lisp files in the Org distribution. Also, it actually never reloaded
any files from the contrib directory. Both of these problems are now
fixed.
Mapping call a function for each matching entry. So far this has
always assumed that the entry stays in the buffer and search can
continue from there. However, when the mapper function removes the
tree, more control is needed to specify from where the search should
continue.
The action function handed to the mapping function can now set the
variable `org-map-continue-from' to the position from where mapping
should continue.
Daniel Hochheimer writes:
> It seems there is a bug in the handling of simple dependencies.
> I think an example tree is the best solution, to show you the bug:
>
> * Projects
> #+CATEGORY: Projects
> *** TODO foo bar project
> :PROPERTIES:
> :ORDERED: t
> :END:
> ***** TODO foo subproject :FooSubproject:
> ******* TODO Task 1
> ***** TODO bar subproject :BarSubproject:
> ******* TODO Task 1
>
> This is in my .emacs file:
> (setq org-enforce-todo-dependencies t)
> (setq org-agenda-dim-blocked-tasks 'invisible)
> (setq org-odd-levels-only t)
>
> the expected global todo agenda view imho is:
>
> Projects: Task 1 :FooSubproject:
>
> but actual it is unfortunately:
>
> Projects: Task 1 :FooSubproject:
> Projects: Task 1 :BarSubproject:
>
>
> Imho "Task 1" from "bar subproject" should not be visible,
> because "bar subproject " is blocked because of the
> ORDERED property (therefore it's childs should be blocked, too)
>
>
> Is it easy / possible to fix this bug? My whole GTD system is
> heavily based on such project / subproject-Constructs. But with
> this bug my global todo agenda view is unfortunately "polluted"
> a little bit with tasks from projects that shouldn't be active.
After some back and forth, Daniel convinced me, and this is now done
correctly.
If the trigger for a log mode entry in the agenda has notes, for
example a note associated with a state change or with a clock entry,
the first line of the notes will now be added to the logbook entry.
You can turn this off the with new variable
`org-agenda-log-mode-add-notes'.
The annotation and initial contents for a remember template are
normally taken from the variables `annotation' and `initial', which
are bound by remember. We now also check the property list for such
values, so that the link generating routine can force the right values
in there.
With the setting
(setq org-refile-use-outline-path 'file)
the file names ended up twice, like
"xxx.org/level 1/level 2 (xxx.org)"
Now the second occurrence is omitted.
During secondary agenda filtering, pressing "?" now will install a
filter that selects entries which do not have an effort defined.
This new model was necessary because we needed to stop interpreting
entries with no effort defines as 0 effort. This was inconsistent,
because for normal agenda sorting, the treatment of these entries
depends on the variable `org-sort-agenda-noeffort-is-high'. Now this
variable is also respected during filtering.
Rustom Mody writes:
> The last two lines of my org file are
>
> *** Vishnu Sahasranam
> *** Ram Navami
>
> without a newline at the end
>
> Trying to reorder these two lines I do a M-S-down on second last
> line I get
>
> *** Ram Navami*** Vishnu Sahasranam
