David Kastrup [Sun, 24 Nov 2013 23:59:55 +0000 (00:59 +0100)]
Issue 3675: Make convert-ly -d only ever update on changed files
Previously, it updated unconditionally whenever a new stable version
came out, leading to merge conflicts. When the final applied
conversion is to an unstable version and the following stable version
is not beyond the conversion target, the following stable version is
used.
Note that this rule does not make a factual difference for continuous
updates of a code base (the normal use case for
scripts/auxiliar/update-with-convert-ly.sh), but it makes a difference
for the conversion/import of code that may have fallen behind a lot
(like with the LSR import, or when converting archived files).
Devon Schudy [Fri, 29 Nov 2013 03:05:13 +0000 (19:05 -0800)]
Grace notes: only shorten previous note if overlapping
Previously a grace note after a rest shortened the last note before the
rest.
This patch shortens the previous note to the start of the grace note
rather than by the length of the grace note, so notes before rests
aren't affected unless the grace notes are longer than the rest.
Devon Schudy [Fri, 29 Nov 2013 03:00:02 +0000 (19:00 -0800)]
Fix spurious error playing initial grace notes (issue 1412)
This happens because the MIDI output starts at tick 0, regardless of
when the first event is. Initial grace notes start start at negative
ticks, so midi_walker::output_event thinks they're out of order. Fixed
by starting at the first event.
David Kastrup [Mon, 25 Nov 2013 18:33:34 +0000 (19:33 +0100)]
Issue 3676: Bar checks display the wrong location
This one is just embarrassing. The lexer used last_input_ where it
should have been using here_input (), resulting in music identifiers
and some other stuff to be located one token early.
David Kastrup [Sun, 17 Nov 2013 18:29:06 +0000 (19:29 +0100)]
Issue 3663: Crash with \repeat ... \alternative and \remove "Bar_engraver"
This approach is just poking around in the dark and removing Scheme
error messages. In particular substituting the original glyph for
left-bar-line and right-bar-line (when missing) in calls to
span-bar::compound-bar-line seems fishy, and the original logic
for setting left-bar-broken if left-bar-line is not even present
is also not clear to me.
The results for the example in the bug report:
form = \new Staff \with {
\remove "Bar_engraver"
} \repeat volta 2 {
s1
} \alternative {
s1 % first ending
s1 % second ending
} % form
\score {
<<
\form
>>
} % score
are also not identical with the 2.16 results: the volta brackets now
run into each other; before there was a gap.
It fixes the crash. I have no idea what a proper fix would look like.
David Kastrup [Mon, 18 Nov 2013 11:37:59 +0000 (12:37 +0100)]
Issue 3668: Let NullVoice have an existence in MIDI
Issue 3457 omitted to introduce the NullVoice context into MIDI. This
implementation is incomplete (see the TODO in performer-init.ly), but
at least it does not let the context structure go off the deep end in
MIDI.
David Kastrup [Sat, 16 Nov 2013 12:24:11 +0000 (13:24 +0100)]
Let a PITCH_IDENTIFIER optionally be followed by octave modifiers
This makes pitch identifiers more similar to note names. There is no
point in keeping the distinction just for triggering syntax errors
on music that has no sensible other interpretation.
"unpitched" simply means that this is rendered as a normal NoteEvent,
but the pitch information is missing and has to be filled in by some
other means. The example using RhythmicStaff is not actually
commendable as the MIDI rendition still requires a pitch or drum
type. It's just that as of this patch, there are no immediately
useful applications. The rhythmic shorthand { c4 4. 8 4 } is
introduced by a later patch.
David Kastrup [Tue, 5 Nov 2013 16:12:17 +0000 (17:12 +0100)]
Let #{ 4 #} be a duration while #{ 4 \cm #} stays a length
An UNSIGNED not followed by a NUMERIC_IDENTIFIER inside of #{ ... #}
is interpreted as a duration rather than a number. Also, inside
of #{ ... #} no "numeric expressions" using binary operators +-*/ are
interpreted.
The behavior of (signed and unsigned) integers and numeric expressions
in assignments and as function arguments rather than inside
of #{ ... #} is not affected.
David Kastrup [Sun, 3 Nov 2013 07:54:21 +0000 (08:54 +0100)]
Issue 3644: A \score-lines markup list command for multi-lines embedded scores
Originally requested as issue 1334.
Like the \score markup, \score-lines is not called through the normal
markup list command mechanisms as it would require something awkward
along the lines of
\score-lines ##{ \score { ... } #}
to get this through. Instead, a reserved word \score-lines in the
parser will manually call the command with an embedded score, leading
to just
\score-lines { ... }
This is somewhat clumsy in several ways: it requires an additional
reserved identifier (no other reserved identifier contains a dash),
and it means that a score may occur in a file without an explicit
enclosing \score command.
Nevertheless, the previous semantics implemented as issue 1334 were
not tenable: particularly after issue 3270 had been passed, it was
close to impossible for the average user to come up with a way of
getting an actual markup list from a score.
Note that there are as of yet no user-level commands spacing a markup
list in a way similar to the spacing inside of a \score markup.
David Kastrup [Sun, 3 Nov 2013 14:30:53 +0000 (15:30 +0100)]
Issue 3645: Allow \partial to occur in mid-piece
This is done by moving measurePosition to negative values when at the
beginning of a measure, and to a respective distance before the end of
the measure in case the measure has already been started.
David Kastrup [Thu, 31 Oct 2013 11:03:02 +0000 (12:03 +0100)]
Issue 2379: resetting autoBeaming after a cadenza
Letting measurePosition track the grace state of current_moment
unconditionally keeps it from moving into uncharted territory when
Timing.timing gets switched on or off during grace times.
Mike Solomon [Sat, 2 Nov 2013 15:35:44 +0000 (16:35 +0100)]
Issue 3631: 2.17 does a worse job with vertical spacing and/or the page layout than 2.16
Looks for prebroken pieces of dead items in the pure relevant function.
Even if the item is dead, its prebroken pieces may not be. We need to check
them and build them into the pure skylines of axis groups used by the
align interface.
David Kastrup [Thu, 31 Oct 2013 22:34:12 +0000 (23:34 +0100)]
Issue 3641: Keep only one Axis_group_engraver active
The internal context property axisEngraver is used for tracking the
currently active Axis_group_engraver. This supercedes the solution
attempted in issue 2990.