activeBranches.txt

gregc/moreMusicSiteFixes:
    Initial tie work is here.  The rest of this has been merged into main v4.0.0.

main (v4.0.0):
    Add ABC import/export (including Opus support)
    Humdrum/MEI import/export: 
        add Opus support (three options with MEI)
        Support styled lyrics
        Better color support
    All: support PedalMarks
    Humdrum import: stem direction fixes
    Merged gregc/spannerAnchorFixes:
        Fix MEI export of SpannerAnchors, so it doesn't cause gaps to be exported.

main (v3.5.0):
    Merged gregc/moreNotation:
        Support RehearsalMarks in Humdrum/MEI import/export.
        Support PedalMarks (disabled until they are in music21)

    Merged gregc/betterXmlIds (fixes for Philippe):
        Better xml:ids (don't prefix existing strings)
        Don't get confused by empty measures during MEI export (don't use offsetIterator)
        Ignore crazy beams during export (number > makes sense for duration)

    Merged gregc/OMREvaluationPaper (see *** MERGED TO develop HERE *** above)
        Humdrum import/export supports .kern, not just .krn.
        Any parseError is stashed in score.c21_parse_err.  If accepting syntax
            errors, and an exception is raised, eat the exception, return an
            empty score, and stash the exception reason in score.c21_parse_err.
        Fix a bunch of Humdrum syntax errors (if requested).
        Allow (ignore) a final line containing '<eos>'.
        New tests:
            parseFolder: parses a folder full of gt and pred files, fixing syntax
                errors in the pred files.  Prints out any parse errors,
                raised exceptions, empty scores, or scores with no staves.
            fixSMB: parses all the ground truth data (from metadata.jsonl) and
                reports which ones have what problems, and a numeric summary
                at the end.

    Merged gregc/OptionToFixBadHumdrumInput:
        An option (for OMR folk using musicdiff) to request that instead of crashing, we do our best
            to fix up the Humdrum file we are importing, so we end up with a score.  I expect that this
            will be triggered by a musicdiff command line option on one of the input files (not
            the ground-truth file, the one that came out of the model)
        Also, musicdiff now handles comparing scores with different numbers of parts.  Stop avoiding that
            case in diffall* and showdiff* tests.

    Merged gregc/moreMusicSiteFixes, but without any of the initial tie work.
    (That is still on  gregc/moreMusicSiteFixes.)
    Here's the merge:
        Factor out some MEI stuff into M21Utilities so MusicEngine can use it.
        MEI export: no ties in chord symbols! (etc)
        MEI export: Treat barline type 'heavy-light' as 'regular' (hopefully it will be combined with
            something else, like a repeat start).  MEI has no heavy-light barline (SMUFL does, and
            Humdrum does, and MusicXML does).
        MEI export: Lots more xml:ids.  Export any string voice.id (overridden by
            voice.c21_label) in layer@label, and make layer@n always be a low integer.
        MEI export/Humdrum export: Make note of all custom m21 attributes we create, and remove
            them when we're done, so we can export multiple times from the same (perhaps modified)
            score without getting confused by things like clef.mei_handled.


main(tag=v3.2.0):
    from gregc/textOutput:
        Tests: changes to match new musicdiff functionality (new details, text output, and all new
            measure-wide-lyrics comparison).
        MEI import/export: fixed a couple slur bugs, one of which (export) was about slurs that
            weren't in the score spannerBundle.
        MEI export, Humdrum export: fixed the rest of the places where we were using spanners that
            weren't in the score spannerBundle.
        MEI export, Humdrum export: Use new M21Utilities.isIn(obj, spannerBundle) for speed and
            accuracy (it does "is" instead of "==")
        ... SlurFixes tests ran here
        MEI import: handle open-ended slurs (no endid, or no startid)
        MEI import: fix import of slur of type: @slur="i1"/"t1"
        Humdrum export: do export slurs that start or end on a rest
        ... MoreSlurFixes tests ran here
        MEI import: support slur placement ('above' and 'below')
        Humdrum import: fix double-slur-on-chord code that puts one 'above' and one 'below'
        MEI import/export: it's slur@curvedir, not slur@place.
        ... EvenMoreSlurFixes tests ran here
        MEI import: convert tstamp and tstamp2 beats from n.nnnnn -> Fraction (using
            DENOM_LIMIT == 768)
        MEI import: fixed some missing opFracs in conversion from offset to beat, that were screwing
            up our newly accurate tstamp/tstamp2 code.

    Humdrum import of lyrics: support *edit/*Xedit and *italic/*Xitalic as synonyms of *ij/*Xij.

    Tests: go back to comparing rest positions with Verovio (notice a few longstanding errors where
        positioned rests are under ottavas).

    from gregc/moreFixesForMusicSite:
        MEI export: better <harmony> text, including use of cs.chordKindStr if present.
        All export: make up chordKindStr if necessary.
        Humdrum import: stop duplicating lyrics across voices (grab code from Verovio)
        All import: simplify chord symbols (try to find better chordKind that has fewer
            chordStepModifications)
        Utilities: chord simplification can now do '7+ add 9' -> '9#5', 'maj7+ add 9' -> 'maj9#5',
            and 'm7 alter b5' -> 'ø7'.
        MEI export: simplify chord symbols to make better harte and better <harm> text.
        MEI export and made-up chordKindStr: use '°' instead of 'dim', in 'o7', etc.
        Humdrum export: make up chordKindStr even if kind=='major' (if chordKindStr == '')
        All export: export poet as lyricist.
        Humdrum export: perform "n=" incrementation even if one of the LO:DY, LO:HP, or LO:H
            instances doesn't have a param that needs "n=".

    from gregc/foundByNewDiff:
        - Fix Humdrum export of numbered layout for note in chord (!LO:N:n=2, not !LO:n=2:N)
        - Fix Humdrum import of Unpitched (lots of cleanup)
        - Humdrum import: be more consistent about stemDirection
        - Humdrum import: better support for ottavas affecting more than one voice
            - Reorder multi-voice ottavas so last element ends at maximum ottava offset and
                first element starts at minimum ottava offset.
            - Don't start an ottava twice (it will nuke the first one).
            - When processing a subsequent voice in the first measure under an ottava,
                pay attention to the ottava's start offset in that measure (don't just
                put all the voice's notes in the ottava).
            - Don't stop ottava too early (wait for the last voice to notice the ottava end)
            - Deal with manipulators that hide voices from ottava start (split just after)
                and stop (merge just before).

    from gregc/chordSymbol:
        - Humdrum import/export of ChordSymbols (read **mxhm and **harte, write **harte).
        - MEI import/export of ChordSymbols (put harte in harm@type for now; hopefully
            MEIv6 or v7 will have something better).

main(tag=v3.1.0):
    ... and so much more...
    Issues discovered during new MusicXML -> MEI conversion tests.
    - MEI export: no implicit text emitted in <tempo>.
    - MEI export: no more duplicated clef/keysig/metersig; if you emit in staffdef, don't emit
        again later.
    - Humdrum import/export: OMD vs Tempo fixes (picked up some import fixes from Verovio, then
        started depending on them in Humdrum export)
    - MEI import/Humdrum import: fix 'title' vs. 'movementName'
    - Humdrum import/export: support non-note-associated dynamics/directions/tempos by adding a voice
        full of invisible rests, one of which lands at the desired offset.
    - MEI import: fix timestamp computation (meter changes inside layers)
    - Modify all diffall and showdiff tests to check number of parts before diffing
        to avoid musicdiff crash.
    - Fix MEI export crash (@midi.bpm must be exported as a string)
    - MEI export/Humdrum export: handle "play only" metronome marks.
    - Humdrum import: fix to handling of transposing instruments
    - MEI export: if @tstamp2 is going to be in a previous measure, don't crash.
    - MEI export/Humdrum export: handle Grace vs Appoggiatura more carefully
    - MEI export/Humdrum export: handle gaps by exporting invisible rests
    - diffall tests: visualize note offset differences (new musicdiff feature)
    - MEI export/Humdrum export: obey chordSymbol.writeAsChord during export (we
        do not export the ChordSymbol at all (yet) if .writeAsChord is False)
    - MEI export/Humdrum export: better handling of staves not in any staffGroup
    - Humdrum export: Regularize treatment of absoluteY
    - MEI import/export: Support lyric placement
    - MEI export: improve absoluteY <-> placement relationship
    - Humdrum import: better handling of preferred vs forced placement (of dynamics/hairpins)

    Random Bugfixes:
    - MEI import/export: DatePrimitive range edge that is also a range needs to be
        "[date1..date2]", not "date1/date2".  We can import both, but export
        always produces "[date1..date2]".
    - MEI export: Add missing @isodate for "datePublished".
    - All: Regularize invisible vs regular vs missing m21 Barlines.
    - MEI import/export: handle tuplet.bracket and visibility better.
    - MEI export: actually emit clef/key/timesig changes that aren't at start of measure.
    - MEI export: handle spanners that start and end in different Parts (we were crashing).
    - MEI export: emit ornament accidentals only if visible (there is no @accid.ges on
        ornamentals, unfortunately, or we would use that for invisible accidentals).
    - MEI import: allow <space> to be used as a @startid or @endid in a staff item (e.g. <hairpin>).
    - MEI export: MEI hairpins end at start of @endid.  Unfortunately, music21 DynamicWedges
        end at _end_ of dw.getLast().  So we can't use dw.getLast() as @endid.  Instead
        we force a @tstamp2 at the ending offset of dw.getLast().

main (tag=v3.0.1):

    Fix MEI metadata export: <appInfo> goes before <editorialDesc> in <encodingDesc>.

    Picked up a few small fixes from latest version of humlib.

main (tag=v3.0.0)

  Export to MEI: handle tuplets that can't nest properly with the voice's beams by emitting
    <tupletSpan> instead of <tuplet>.

  Export to MEI: export visual+gestural duration @dots.ges non-ambiguously
    (i.e. always add @dots.ges, even if it's zero or equal to @dots).

  Rework Humdrum export of tuplets, including carefully choosing non-power-of-two recip.

  MEI import: Improved support for gestural duration, including support for @dur.ppq (which is always gestural).

  Import and export MEI metadata

  Import MEI v4 and v5 (decided via root element @meiversion)
  Export MEI v4 and v5 (default to v5, do v4 if MeiWriter.meiversion.startswith('4'))

  Humdrum export: Ottavas
  Humdrum export: Articulation placement

  MEI exporter work (and associated improvements to other importers/exporters)
  Rename of diffall*.py and showdiff*.py (with addition of MEI export variants)

  Humdrum import/export: Support center justification (MEI import already worked)

  MEI import: Support <dir>, <dynam>, <tempo>, <hairpin> with @startid instead of @tstamp
    and/or @endid instead of @tstamp2.

  meiConverterClass:
    1. Move all the global functions into the converter
    2. Move otherInfo into the converter instance storage
    3. Get rid of self.otherInfo in favor of converter instance member data

  meiStaffGroups:
    MEI import: support <staffGrp> properly -> multiple PartStaffs and a StaffGroup
        - lots of stuff including braces/brackets, labels/labelAbbrs, etc
        - make sure voice numbers are unique across staff groups
        - BOTH importers should do the automatic detection of piano(2)/organ(3) scores. Humdrum already
            does, but it's a bit too aggressive.  It should check for non-keyboard instruments before
            doing it.  MEI should add this code (and do the same instrument/part label check).
    Humdrum import: several StaffGroup fixes.
    tests/diffall*.py: visualize the new staff group differences

  Add a runtime check in converter21.register() that raises if music21 v9.0.0a11+ is not present.

  python3.10 -
    Require python3.10 and music21 v9.
    Clean up typing stuff by using Blah | Bleah | None instead of t.Optional and t.Union, etc.
    Replace t.List, t.Tuple, t.Set, t.Dict with list, tuple, set, dict.
    Remove all checks for music21 features, since they are all there.
    Update copyright dates to 2023
    Update converter21 version to 3.0.0

  All: Support ornamentAccidentals

  MEI import: use @dur.ges if @dur is missing.  If both are missing assume dur=None as before.

  tupletBug - pick up fix for https://github.com/humdrum-tools/verovio-humdrum-viewer/issues/697, a.k.a. fix 7tupletsOf16thNotes to actually have 16th notes instead of 32nd notes.  Fixes a TON of beam differences in diffallmei.py.

  ottavaTransposeFix

  delayedTurns

  tupletDetails