windfisch/muse

My personal repository of the MusE sequencer

TESTING

muse's Issues

setting a wave track to LOCKED mode and then back to UNLOCKED breaks things

when altering the playing tempo (decreasing it, iirc) the audio starts to "run away".

poslen needs TIME mode

drop FRAME, but replace by TIME mode.

disallow cloning then :/!

must (!) store PartLists, EventLists and possibly more with XTick, not unsigned, as key!

Cleanup XTicks implementation

Currently, they're hacked to fit into the rest. MusE needs some rework to support them "natively".

audio glitches when seeking

I guess that the stretcher still has old data in the buffer. This needs to be emptied first, and there might be a need for lag compensation.

check accuracy, lag compensation

Testing is needed whether the audio stream lags, or is time-accurate

wavecanvas (and possibly more places) die in Event.setFrame due to mismatched posType

gui for wavefile's tempomap

importing a wave to a non-timelocked part does not enable stretch

one must lock, then unlock again to get stretching :/

markers in tick or time mode

load/save stuff to files

Add to the file format:

• stretch mode
• file tempomap (also to an extra file?)
• XTicks

wave parts passthrough

wave parts which contain only one event shall offer the event's functions directly from the partcanvas

support stretcher quality settings

Rubberband and Samplerate support quality/speed tradeoffs. Support them, maybe in a generic way.

Selected parts do not enable Edit->Copy action

In the audiomsg_overhaul branch, selecting a part does not enable Edit->Copy nor Ctrl+C.

Also: in audiostreams (Todo: test whether in master) creating a part (which is then selected automatically) does not enable the action. Selecting it by hand, however, does.

paste dialog crashes, when a wave is copied.

Reproduce:

Create 2 wave tracks, with events on them.
Select on wave event, copy it.

Then Functions -> Paste dialog.
MusE will crash, relevant backtrace is:

(gdb) bt
#0  0xb75a3725 in MusECore::Pos::xtick (this=0x4) at /home/flo/muse-git/muse2/muse/pos.cpp:275
#1  0xb75ef386 in MusECore::WaveEventBase::setAudioFile (this=0x90aa5c8, path=...) at /home/flo/muse-git/muse2/muse/waveevent.cpp:121
#2  0xb75ef64e in MusECore::WaveEventBase::read (this=0x90aa5c8, xml=...) at /home/flo/muse-git/muse2/muse/waveevent.cpp:159
#3  0xb75388e5 in MusECore::Event::read (this=0xbfbac59c, xml=...) at /home/flo/muse-git/muse2/muse/event.cpp:156
#4  0xb75c0c40 in MusECore::readXmlPart (xml=..., track=0x92377f8, doClone=false, toTrack=false) at /home/flo/muse-git/muse2/muse/songfile.cpp:336
#5  0xb6f3a89b in MusECore::get_paste_len () at /home/flo/muse-git/muse2/muse/widgets/utils.cpp:535
#6  0xb7306969 in MusEGui::PartCanvas::cmd (this=0x8a8e628, cmd=7) at /home/flo/muse-git/muse2/muse/arranger/pcanvas.cpp:2712

rewind and stop-then-play freezes

Tested on my slow Thinkpad. MusE seems to hang a bity then continues working

audioprefetch sync/locking issues

see muse-sequencer#339

Put (wave-)event preparation/"defusing" into executeOpGroup[13]

Event, especially WaveEvent needs to create an AudioStream before inserting into a part. However, when altering the Event (e.g. moving it), it is duplicated, then changed.
And with it, the AudioStream and all resources tied to it are duplicated! (Plus, that happens in various contexts, which is bad to maintain)

Instead, Events (and possibly more items in MusE) need an "arm()" and a "defuse()" function, which is called from and ONLY from the execute/revertOperationGroup1/3() functions (i.e. those which are not executed in realtime context).

remove "startXtick" from audiostream: can access it via EventBase functions

with latest commits, EventBase knows about its absolute position. No need for audiostreams to store this as well

Breakage when using the global tempo manipulator

MusE seems to freeze/cause 100% cpu usage then.

Pos::frame fixes

must not be mutable
set it on every change instead. upon tempo change, iterate through them all. (???)

gui for choosing a stretch mode

Popup menu for WaveEvents or for WaveParts which only contain a single Event to select the stretching mode and quality.

raster points for parts

While a part might start somewhere, its beginnig might not align with a "beat".

Instead, the user shall be able to set a marker somewhere inside the part, letting the part snap not relative to its beginning but relative to this marker.

time-lock wave parts

partcanvas should draw time-locked parts correctly!

enable both Tick and Frame editor in MarkerView

make sure that the tick editor does not "snap", losing accuracy

cue points

"TIME" - to - TICK sync points. auto-adapt tempo so that the given tick ALWAYS happens at the required time

Must disable/reenable audiostreams when moving to a different track (channels() may change)

is moving a played event supported?

Does MusE misbehave when moving a WaveEvent/Part that is currently played back? (It shouldn't)

Multiprocessing for fetchAudio

this can be done independently from audiostreams.

Multiprocessing in general is pretty hard to realize; however, parallelizing only the prefetcher should be easy.

Clones don't load their clone-ness from a file

After loading a file, parts are not cloned any more.
A file saved by audiomsg_overhaul is correctly loaded by current master.
A file saved by current master (or audiomsg_overhaul) is NOT loaded correctly by audiomsg_overhaul.

Also check for copy'n'paste

in audiomsg_overhaul: SelectEvent/SelectPart can be bypassed!

CItems have a nonconst Event smart pointer, that can be (and IS) used for using ->setSelected on the CItem, and thus on the Event.
This bypasses the Select{Event,Part} mechanism, leading to confusion and hard-to-find bugs!

waveform graphic for second audio channel contains garbage

Steps to reproduce:

create wave track
open click.wav
(maybe move around?)
enlarge the row, so you see both channels
switch around between timelock and not.

result: the bottom channel will contain random garbage

use wave event's tempomap

a menu item which overwrites the currenr tempomap with the tempo.of the waveevent.

plus: "the same, but make it begin/end with the current tempo"