tick 0)

audioPort as nullptr by default.

frames by the masterGain.

support.

SampleTCOs.

"played" (recorded).

SampleBuffer.

That solves problem with recording that caused samples to be moved and
messed up.

Thanks @-BaraMGB for helping.

Previously, begining record in any position would result the data being
in the end of the TCO.

@see LMMS#3947 (comment)

Thanks to @-BaraMGB :)

`updateMenu`, use a virtual function overriden by every Track
implemetation.

input channel configurations. And support recording of one channel and
converting it to two.

The user would be able to set their input config per track, or globally.
For each recorded sampletrack, the recorded will get the channel config
and will duplicate the samples from one channel to the other in a case
of mono input.

options unchecked will be considered as global default.

Suggested by @-tresf, Thanks!

actual data starts after that.

instead of hacking sampleBuffer ()->startFrame ().

That solves a bug with `startFrame ()` being negetive in recording some
cases.

rate.

The deadblock was caused by the mixer stopping which make the recording
thread to wait until it get started again and in the other side, the
exporting thread which wants to acquire a mutex that had been locked by
the recording thread.

Many thanks to PhysSong :)

support is not available.

Thanks to @-Sawuare.

played correctly after saving and loading a project due to sample rate
not getting saved in the project file.

right place when it not played from the begining.

That has created a difference between the ticks and the metronome and
the sample track.

The cause of the problem was that the calculation of the frame to play
was wrong: we had calculated `framesPerTick` according to the current
engine's sample rate instead of the SampleBuffer's sample rate.

automaticlly create "record sample tco" if there is no recording tco.

This commit adds per-track record setting. If a sample track is set as
record, lmms will check if there is a tco that set to record:
  * If there is, it will record into it.
  * If there is not, it will create a special tco that get created on
recording, and as long as the recording.

and SampleTrack::m_recordingChannel.

avoid a warning.

Thanks to @-PhysSong.

Many thanks to @-PhysSong.

recording would stop early

of writing "Rec").

in order to make stantard containers be able to store it.

addData.

visualizing of a SampleTCO.

SampleRecordHandle: Instead of adding buffers to a list and then joining
it to new SampleBuffer. Add the in-recording to the new SampleBuffer.

SampleTCOView: Make sure our SampleBuffer is not locked, if it is, skip
this frame and flag it for update.

Also make sure we're getting updated for a changes from
SampleBuffer::sampleUpdated.

See the HACK comment

resetData. That solves a race condition when one thread calls
sampleBuffer() and it gets destructed by another thread that calls
setSampleBuffer.

number of functions:
- beginBufferChange, doneBufferChange: Guards for direct access to
m_data. They will also resample the data if it is needed. Functions like
`addData` will check additional parameter to know if we should lock the
mixer in this context.
- Add reverse and remove the attribute m_reversed: The only user was
AudioFileProcessor and it added complexity for the file loading process.
Right now we'll just `reverse` the buffer manually.
- changeAudioFile: Will reset the current buffer and load a file into
it.
- Remove the raw constructor `SampleBuffer(sampleFrame*, f_cnt_t)`: It
was used only once and could easily be converted to another constructor.

isFinished (when possible).

That will solve our "never ending recording after tco finishing" problem
:)

loops.

as not record.

1. Resampling was the wrong (src_rate was invalid)
2. SampleBuffer was using baseSampleRate as the default samplerate
instead of the actual processingSampleRate.

Improve performance by using std::vector::resize instead
of deleting and reallocating every time.

LmmsBasics: Add static asserts to make sure no padding is
added to sampleFrame by std::array.

Thanks to @-lukas-w!

Don't use sharedObject. It does not make any sense and could lead to
bugs when allocated locally (see audio_file_processor). When needed,
this commit used std::shared_ptr which was made for that.

Exporting and restoring is universally done now by saveSettings and
loadSettings. That's allows us to make sure we don't left any important
attributes like "sampleRate" behind

Mixer's processing rate.

That could lead to ireversable sample loss. Now we'll only resample when
it needed and the processing sample rate is higher than the buffer's
sample rate.

Thanks to @-PhySong for pointing it out!

possible.

issues.

making any changes in the track.

sessions and handle sample start offset properly.

totalFrames.

Many thanks to @-PhySong. What a great find.

channel > 2.

resetData.

I've added two new methods:
	* tryAddData
	* tryResetData

Which do the same, but without locking. If we've failed to lock, no
problem, just wait until the next time.

Fixes a regression from 7bc0056 that
wasn't fixed properly in c8337dc.

Fixes regression from f765773 (presumably
a copy/paste mistake)

Thanks to @-PhysSong.

Fixes a regression in 84a1f35.

Fixes a regression in 7bc0056
that breaks showing file names.

* Enable WeakJACK only if JACK is available.

* Try `pkg_check_modules` first when using WeakJACK, too.

Fixes a regression in 063b7c9.

* Bundle 64-bit version of JACK for 64-bit installer.
* Change some CMake logic for NSIS

should return with empty buffer.

…ocked.

Since it may cause quality loss, we have decided to remove it and just
use on-the-fly resampling while playing.

Also renames misleading `shouldLockMixer` to `syncWithMixer`.

requestChangeInModel.

Deprecate direct calls to requestChangeInModel.

- SampleBufferData: Responsible for audio data.
- SampleBufferFileHelper: Responsible for loading audio files.
- SampleBufferPlayInfo: Responsible for holding information such as
startFrame, etc.

This commit also removes `m_varLock` and guards all
getters and setters with RequestChangesGuard. It
will be replaced in the next commit with:
   - Setters: The same but async.
   - Getters: signal on change.

QtConcurrent.

while exiting.

This commit renames createTCO to unsafeCreateTCO and
add a mixer guard on the new createTCO.

and add UpdatingVariable for SampleBuffer's info.