Knife Tool for Sample Clips (Again)

[Spekular]

New version of #4998, with copy constructors for SampleBuffer and SampleTCO and a scalpel cursor when hovering over sample tracks in Knife/Split mode. Also adds copy assignment operator for SampleBuffer and a comment about this being missing in SampleTCO.h.

It's worth noting that this PR currently creates a new sample buffer with each cut, because apparently sample clips rely on having a unique buffer. This means that memory use will increase by the samplebuffer size with each cut. Cloning clips currently behaves the same way, so this isn't a regression, but I believe it eventually needs to be fixed as this leads to performance issues.

TODO:

• Add a TODO comment about the sample buffer issue

[LmmsBot]

🤖 Hey, I'm @LmmsBot from github.com/lmms/bot and I made downloads for this pull request, click me to make them magically appear! 🎩

Linux

• Linux (AppImage): lmms-1.3.0-alpha.1.101%2Bg572b6a8-linux-x86_64.AppImage (build link)

Windows

• Windows 32-bit: lmms-1.3.0-alpha.1.101%2Bg572b6a8d5-mingw-win32.exe (build link)
• Windows 64-bit: lmms-1.3.0-alpha.1.101%2Bg572b6a8d5-mingw-win64.exe (build link)
• Windows 32-bit: lmms-1.3.0-alpha-msvc2017-win32.exe (build link)
• Windows 64-bit: lmms-1.3.0-alpha-msvc2017-win64.exe (build link)

macOS

• macOS : lmms-1.3.0-alpha.1.101%2Bg572b6a8d5-mac10.14.dmg (build link)

🤖

{"platform_name_to_artifacts": {"Linux": [{"artifact": {"title": {"title": "(AppImage)", "platform_name": "Linux"}, "link": {"link": "https://12522-15778896-gh.circle-artifacts.com/0/lmms-1.3.0-alpha.1.101%2Bg572b6a8-linux-x86_64.AppImage"}}, "build_link": "https://circleci.com/gh/LMMS/lmms/12522?utm_campaign=vcs-integration-link&utm_medium=referral&utm_source=github-build-link"}], "Windows": [{"artifact": {"title": {"title": "32-bit", "platform_name": "Windows"}, "link": {"link": "https://12525-15778896-gh.circle-artifacts.com/0/lmms-1.3.0-alpha.1.101%2Bg572b6a8d5-mingw-win32.exe"}}, "build_link": "https://circleci.com/gh/LMMS/lmms/12525?utm_campaign=vcs-integration-link&utm_medium=referral&utm_source=github-build-link"}, {"artifact": {"title": {"title": "64-bit", "platform_name": "Windows"}, "link": {"link": "https://12523-15778896-gh.circle-artifacts.com/0/lmms-1.3.0-alpha.1.101%2Bg572b6a8d5-mingw-win64.exe"}}, "build_link": "https://circleci.com/gh/LMMS/lmms/12523?utm_campaign=vcs-integration-link&utm_medium=referral&utm_source=github-build-link"}, {"artifact": {"title": {"title": "32-bit", "platform_name": "Windows"}, "link": {"link": "https://ci.appveyor.com/api/buildjobs/l47dfyj97k2o0epx/artifacts/build/lmms-1.3.0-alpha-msvc2017-win32.exe"}}, "build_link": "https://ci.appveyor.com/project/Lukas-W/lmms/builds/37736706"}, {"artifact": {"title": {"title": "64-bit", "platform_name": "Windows"}, "link": {"link": "https://ci.appveyor.com/api/buildjobs/dwxivln8okg8jmnm/artifacts/build/lmms-1.3.0-alpha-msvc2017-win64.exe"}}, "build_link": "https://ci.appveyor.com/project/Lukas-W/lmms/builds/37736706"}], "macOS": [{"artifact": {"title": {"title": "", "platform_name": "macOS"}, "link": {"link": "https://12521-15778896-gh.circle-artifacts.com/0/lmms-1.3.0-alpha.1.101%2Bg572b6a8d5-mac10.14.dmg"}}, "build_link": "https://circleci.com/gh/LMMS/lmms/12521?utm_campaign=vcs-integration-link&utm_medium=referral&utm_source=github-build-link"}]}, "commit_sha": "4c604d86dc8cb3b49bdb590639b4a1aca413f333"}

[thmueller64]

Tested, works great!

[Spekular]

Issues with move and resize should be fixed. While I was at it I ended up making box select behave in an (IMO) better and (IME) more standard way. Rather than toggling over to box select mode when control is held, it just starts a box selection when you control + click an empty space, and doesn't visually affect the selected edit mode. On release the edit mode is restored by re-triggering the selected edit button, which seems to be less buggy than the previous implementation based on my quick tests (no more getting stuck in select mode).

[PhysSong]

Is the stuck Ctrl issue #5394?

[Spekular]

Is the stuck Ctrl issue #5394?

Yes. I can still reproduce that issue in 1.3-alpha.1, but not in the current state of this PR.

[IanCaio]

On TrackContentObjectView::mouseMoveEvent() if there's no action being done the following code runs:

		SampleTCO * sTco = dynamic_cast<SampleTCO*>( m_tco );

		if (sTco && m_trackView->trackContainerView()->knifeMode())
		{ setCursor( m_cursorKnife ); }


		if( ( me->x() > width() - RESIZE_GRIP_WIDTH && !me->buttons() && !m_tco->getAutoResize() )
		||  ( me->x() < RESIZE_GRIP_WIDTH && !me->buttons() && sTco && !m_tco->getAutoResize() ) )
		{
			setCursor( Qt::SizeHorCursor );
		}
		else
		{
			leaveEvent( NULL );
		}

There are two issues I noticed with this block:

1. If you are on the Knife mode there's some "flickering" of cursor changes when hovering the edges, because both if (sTco && m_trackView->trackContainerView()->knifeMode()) and if( ( me->x() > width() - RESIZE_GRIP_WIDTH && !me->buttons() && !m_tco->getAutoResize() ) || ( me->x() < RESIZE_GRIP_WIDTH && !me->buttons() && sTco && !m_tco->getAutoResize() ) ) are true, so you set the cursor to the m_cursorKnife and right after you set the cursor to Qt::SizeHorCursor.

2. If you are on the Knife mode, hovers the mouse on the middle of the sTCO and leave it, switch to the Draw Mode and hover again in the middle of the sTCO you'll end up with the knife cursor. That's because when leaveEvent is called, the cursor of the widget is still m_cursorKnife, which is of type Qt::BitmapCursor. So the conditional that switches the cursor back to the m_cursorHand isn't met and you end up with a knife on the draw mode, until you move to one of the edges and then back to the middle. That's because when you move to the edges the cursor is changed to Qt::SizeHorCursor, then when you move back to the middle the if( cursor()->shape() != Qt::BitmapCursor ) conditional is true and leaveEvent changes the cursor back to the hand one.

To be honest I'm not sure why leaveEvent is being called in that particular part of the code instead of dealing with the cursor changes inside TrackContentObjectView::mouseMoveEvent alone. Maybe someone can clarify if there's another meaning for that method call there. If not, we can get rid of it and just rewrite the logic so it accounts for those situations more appropriately, checking for all 3 situations (mouse in the middle in knife mode, mouse in the middle in draw mode and mouse on the edges).

[IanCaio]

And as mentioned in Discord, there is a small bug I still didn't manage to track down, could use some help on this one. The steps to reproduce it:

1. Add a SampleTCO to a Sample Track
2. Split that SampleTCO
3. Try to Copy and Paste (either through the context menu or by dragging and dropping) the left-most SampleTCO from the two splitted ones

It will fail to paste the TCO, pasting a infinitely small SampleTCO instead. If you move or resize the left-most SampleTCO before copying and pasting however it does work. Which makes me think that it has something to do with some update routine that is not being called and is causing the TCO data to be corrupt or something. I still didn't find out what it is, I only know that copying the right-most SampleTCO works fine.

I also unresolved a change request from a one-liner if that ended up not being changed in the last commits. Besides those issues I don't see anything else. I like the changes on the way Ctrl is being handled on the Song Editor. If I find out anything more about this bug I'll update you.

[IanCaio]

Already messaged on Discord, but just for reference, apparently the cause of the bug mentioned above are the following lines:

lmms/src/gui/TrackContentObjectView.cpp

Lines 606 to 607 in 245160d

	// move or resize
	m_tco->setJournalling( false );

Even though the comment says move or resize, it's being called for Move, Resize, ResizeLeft and Split. However, the journalling is only being set back to true for the first 3:

lmms/src/gui/TrackContentObjectView.cpp

Lines 966 to 970 in 245160d

	else if( m_action == Move || m_action == Resize || m_action == ResizeLeft )
	{
	// TODO: Fix m_tco->setJournalling() consistency
	m_tco->setJournalling( true );
	}

Because setJournalling is not set back to true on the SampleTCO being split, when it's being copied saveState will not save the data from the TCO because it only does so when isJournalling is true. Moving or Resizing the TCO fixes the problem because then mouseReleaseEvent sets the journalling back to true.

[IanCaio]

As discussed on Discord, I pushed the last review's suggestions to the PR, gave the code another broad look and it all looks fine.

Style guidelines kept aside, the code LGTM

[PhysSong]

@DomClark Do you still want to review this?

[DomClark]

Looks good to me. include/Track.h and src/core/Track.cpp only have formatting changes; can you revert these before merging?

[Spekular]

Reverted, and merging now! Huge thanks to everyone who reviewed and helped out, including of course BaraMGB for making the PR this is rooted in.

[JohannesLorenz]

Causes #5949 .