John Mandereau [Fri, 8 Feb 2008 13:25:03 +0000 (14:25 +0100)]
Merge branch 'master' of ssh://jomand@git.sv.gnu.org/srv/git/lilypond
* 'master' of ssh://jomand@git.sv.gnu.org/srv/git/lilypond:
Bump version.
Add test for new ligature bracket engraver.
Fix #83 / #297.
MusicXML: Test files: Pedal marks are no notations, but directions
MusicXML: Small fixes/improvements in the test files
John Mandereau [Thu, 7 Feb 2008 15:32:10 +0000 (16:32 +0100)]
Merge branch 'master' of ssh://jomand@git.sv.gnu.org/srv/git/lilypond
* 'master' of ssh://jomand@git.sv.gnu.org/srv/git/lilypond:
Update from Trevor.
Update from Ralph.
@unnumberedsubsubsec -> @subsubsection.
Update from Trevor.
MusicXML: The dynamics test have been moved to 03a-Directions-MusicXML.xml
MusicXML: Improve test suite for the direction elements
MusicXML: Don't try to convert #text as a dynamics tag!
MusicXML: Better support for dynamics
Another fix for mpost: $(outdir) is relative. Also: chain commands with &&.
MusicXML: Improve test suite for the direction elements
Instead of some exported (incomplete) MusicXML files, use
a hand-crafted test, that really contains all elements
defined in MusicXML, even if they are not supported by
any application yet...
-) Use #(make-dynamic-script...) instead of \markup for unknown dynamics
-) Add definitions of unknown dynamics at the top of the converted file
-) Add test case for note-attached dynamics
MusicXML: Replace the notations tests with a better one
So far, the tests were simply generated in some application and then
exported to MusicXML. Since not everything is supported, the tests were
incomplete, inconsistent and sometimes even wrong.
This new test for all notation elements in the MusicXML spec is hand-crafted
and should be more or less complete...
Otherwise the "a2"/"Solo" markups will collide with slurred notes and not
be shifted. Using #'outside for avoid-slur causes the a2/Solo markups
to be always above the slur, which is most legible.
Signed-off-by: Reinhold Kainhofer <reinhold@kainhofer.com>
I usually start with the general structure of a large score
and only then fill it with music. This means that partcombine
is fed two empty music expressions (or only containing one
multi-measure rest each), so that evs1 and evs2 won't contain
data and (assoc "one" evs1) will be an empty list => cdr on
it will crash.
In this case, simply set the split list to an empty list and
don't crash...
Signed-off-by: Reinhold Kainhofer <reinhold@kainhofer.com>
John Mandereau [Thu, 31 Jan 2008 09:15:56 +0000 (10:15 +0100)]
Keep lilylib.py:encoded_write() compatible with Python 2.4
UTF-8 output encoding may not be desired in all cases, but at least
this behavior is closest to the older one (i.e., calling gettext
instead of ugettext and having all POs encoded in UTF-8).
John Mandereau [Wed, 30 Jan 2008 11:08:46 +0000 (12:08 +0100)]
Merge branch 'master' of ssh://jomand@git.sv.gnu.org/srv/git/lilypond
* 'master' of ssh://jomand@git.sv.gnu.org/srv/git/lilypond:
Oops, fix variables in macro.
More 3-value macro LSR stuff.
Begin using 3-value @lsr macros.
John Mandereau [Wed, 30 Jan 2008 11:07:50 +0000 (12:07 +0100)]
Use Python Unicode encoding for gettexted strings
Python 2.5 strings encoding handling seems not always consistent
between various modules, so we must use Python Unicode strings and
thus reencode strings when writing to stderr/stdout. This commit
applies changes proposed on lilypond-devel a while ago in thread
"Gettexted messages encoding with Python 2.5".
Even if GUB still uses Python 2.4, LilyPond packages on some distros
(e.g. Fedora) already use 2.5, so it's important to fix this. I think
these changes are also compatible with 2.4, but anyway I must check
GUB binary when it's out.
By the way, further localization work is needed: localize musicxml2ly
strings, and fix some nitpicks.
John Mandereau [Wed, 30 Jan 2008 08:55:51 +0000 (09:55 +0100)]
Mark more strings for localization
Half-localized convert-ly messages "Not smart enough to convert..."
look awkward, so I marked a lot of new strings for localization in
convertrules.py.