Include past automation patterns in processing by lukas-w · Pull Request #3382 · LMMS/lmms

lukas-w · 2017-02-23T20:21:30Z

Closes #662.

I think that ideally, automated values should update when setting the playing position, not just when playing. Opinions on that?

#662

tresf · 2017-02-23T20:56:58Z

Holy mackerel! 🎉

zonkmachine · 2017-03-07T10:37:02Z

Tested. 👍

I think that ideally, automated values should update when setting the playing position, not just when playing. Opinions on that?

Probably yes to this but there are other cases where values aren't updated until you press play. For instance when you import a midi file and it's tempo isn't updated until you play.

lukas-w · 2017-03-07T15:37:12Z

Unfortunately this PR causes an issue where some automation patterns are ignored, because automation tracks are processed in the order they appear in the song editor. See the following screenshot, where the second track overwrites the first one:

For instance when you import a midi file and it's tempo isn't updated until you play.

Isn't that just a global automation track on the tempo? Doesn't sound like desired behavior to me.

lukas-w · 2017-03-07T23:30:12Z

some automation patterns are ignored

Fixed via c42eb2e.

I did a quick test implementing "Updating when setting the position", but I can't seem to find a good place in the code to insert the updating procedure at.

lukas-w · 2017-03-09T22:19:54Z

Do we want to include this PR in 1.2?

Umcaruje · 2017-03-09T23:12:03Z

Do we want to include this PR in 1.2?

I really like this feature, but we should be in a feature freeze, so not really sure. Interested in what others have in mind. I'm fine either way really.

tresf · 2017-03-09T23:41:05Z

Do we want to include this PR in 1.2?

It's a pretty easy fix to revert if it causes problems. If someone can provide some (additional) testing, that would certainly help. @musikBear is usually good for this if we can get him a win32 installer.

musikBear · 2017-03-10T10:37:25Z

@lukas-w Marvellous 🎆

musikBear · 2017-03-10T10:40:08Z

@lukas-w Marvellous 🎆
That one is indeed a good feature.
If its decided that it goes past the freeze, and into 1.2, just '@' me a notification for the test-file

jasp00

Some issues to address. Then, this should be merged.

jasp00 · 2017-03-17T18:28:55Z



 	void processNextBuffer();
+	void processAutomations(const TrackList& tracks, MidiTime timeStart, fpp_t frames, int tcoNum);


Method should be private.

I'm going to add a unit test for this feature, so I'll leave it public. We can gladly refactor to split this into a public and a private part some day.

If this is really private, you should not be testing private methods. You should publish when necessary.

you should not be testing private methods

Good thing this one's public then. 😉 As I said, we can refactor this some day, but it's out of scope for this PR.

Why cannot processAutomations() be private until there is a real need to be public? May I do the cut+paste?

jasp00 · 2017-03-17T18:31:25Z

 	}

+	/// Returns true if and only if a->startPosition() < b->startPosition()
+	static bool comparePosition(const TrackContentObject* a, const TrackContentObject* b);


This is only needed inside Song and uses a public TrackContentObject function. It should a static function in Song.cpp.

We may end up using it elsewhere. Also, it's quite obviously a function related to TrackContentObject, so I don't think moving it to a mostly unrelated file is a good idea. Whether TrackContentObject should know it's own position is debatable of course, but that's how it is right now.

I do not agree, but your choice is not bad.

jasp00 · 2017-03-17T18:32:37Z

+			TrackContentObject* tco = track->getTCO(i);
+
+			// Skip this TCO if it's muted or not contained in the time period we're processing
+			if( tco == NULL || tco->isMuted() || tco->startPosition() > time_end)


tco == NULL is an error. It should not be hidden.

I agree, but I didn't write that part, I just moved it and didn't dare remove the check introduced by @tobydox 9 years ago. Feel free to add something like a qCritical message or to just remove the check.

lukas-w · 2017-03-21T15:31:11Z

Added tests and some refactorizations, which also makes processAutomations private, @jasp00.

jasp00 · 2017-03-21T22:06:22Z

Good. Could you add the scope trick for automatedValuesAt()? Something like privateAndTest that is defined to private normally and to public on tests.

lukas-w · 2017-03-22T07:51:41Z

Could you add the scope trick for automatedValuesAt()?

What? Why?

# Conflicts: # include/AutomationPattern.h # src/core/AutomationPattern.cpp # src/core/Song.cpp

lukas-w · 2017-03-22T22:13:04Z

I think it's better to not merge this with 1.2. It may be a low risk, but we should take the term "feature freeze" seriously if we want to have a working 1.2 release any time soon.
I'll create a new stable-1.2 branch, then merge this into master.

jasp00 · 2017-03-23T07:48:33Z

Perhaps you should not have added so many changes to the PR, but this is a bug fix. I do not know why #662 is tagged as "enhancement". After all your work, if you are not going to finish, I will.

lukas-w · 2017-03-23T12:58:54Z

It may be a bug fix from the user perspective, but from the code perspective, it can be considered a rewrite of automation processing. So I suggest not merging this without extensive testing, which is better done in part of the next release so we don't hold up this one even longer.

if you are not going to finish, I will.

What's not finished here?

tresf · 2017-03-23T14:50:29Z

I do not know why #662 is tagged as "enhancement". After all your work, if you are not going to finish, I will.

I think @jasp00's point here (hard to read between the lines) is that if this is a bug, it can be merged as a bug-fix after the freeze... whereas as an enhancement, it would wait until the next release.

In regards to "why" this is an enhancement (I labeled it as such), it is because the automation behavior was behaving correctly, just not intuitively. e.g. the old behavior was not a bug and it was not wrong.

From a delivery perspective, this feature adds a tremendous amount of value to 1.2 so if we could get a precompiled version out for our testers, that could certainly influence the candidacy.

lukas-w · 2017-03-23T15:39:56Z

Here you go: lmms-win32-pr-3382-v1.2.0-rc2-113-ge2808837d.zip

tresf · 2017-03-23T19:12:47Z

Testing feedback from @specular:

@tresf 26 automation tracks and no playback issues. Tested reading backwards on a few and it works. 👌

On a side note, there are some items that are being duplicated, namely all instruments as well as the menu item for LADSPA Plugin browser. I can't see it being related, but wanted to relay that feedback.

lukas-w · 2017-03-26T10:24:42Z

On a side note, there are some items that are being duplicated

This probably surfaced only now because plugin search paths were broken until 7251c84. Fixed via 0d77cef.

So it sounds like no one's objecting to having this in 1.2 (except me). Merging 👍

jasp00 · 2017-03-28T08:44:02Z

Could you add the scope trick for automatedValuesAt()?
What?

privateAndTest:
    static AutomatedValueMap automatedValuesAt(const Track::tcoVector& tcos, MidiTime time);

Why?

Because automatedValuesAt() should not be used by other classes.

Tests should be under tests/src.

lukas-w · 2017-03-28T10:14:02Z

Because automatedValuesAt() should not be used by other classes.

I understand what the private keyword is for. But there's no point in making this function private and introducing a "scope trick" doesn't exactly encourage good design either. It's a static, read-only function. No damage done with this one public.

Tests should be under tests/src.

They are.

jasp00 · 2017-03-28T13:38:27Z

No damage done with this one public.

How does the "scope trick" discourage good design? It observes encapsulation restrictions. If you publish a private method, other classes may depend on implementation details.

Fixes LMMS#662 * Include past automation tracks in processing * Track::getTCOsInRange: Use binary search, fix doc * Automation refactorings * Add automation tests

Include past automation tracks in processing

b2bda9e

#662

lukas-w changed the title ~~Include past automation tracks in processing~~ Include past automation patterns in processing Feb 23, 2017

Process automation tracks globally

c42eb2e

lukas-w force-pushed the iss-662 branch from f559b3e to c42eb2e Compare March 7, 2017 19:15

jasp00 reviewed Mar 17, 2017

View reviewed changes

Javier Serrano Polo and others added 4 commits March 19, 2017 17:58

Do not ignore unexpected error

b0ecb59

Automation: Don't record when play position is outside of pattern

0e5b66f

Track::getTCOsInRange: Use binary search, fix doc

554f901

Automation refactorings & record fix

42fcaed

Add automation tests

b23041b

lukas-w force-pushed the iss-662 branch from 4bce488 to b23041b Compare March 21, 2017 15:49

lukas-w added 2 commits March 22, 2017 09:28

Merge branch 'master' into iss-662

03740fd

# Conflicts: # include/AutomationPattern.h # src/core/AutomationPattern.cpp # src/core/Song.cpp

No C++11 in headers (breaks plugins)

429da92

lukas-w added 2 commits March 23, 2017 18:06

Automation: Fix empty patterns (regression from 42fcaed)

e280883

Automation: Fix processing with tcoNum

54d2126

lukas-w changed the base branch from master to stable-1.2 March 26, 2017 10:24

lukas-w merged commit 6004eda into stable-1.2 Mar 26, 2017

lukas-w deleted the iss-662 branch March 26, 2017 10:27

lukas-w mentioned this pull request Mar 26, 2017

Make automation read back to the last automation's clip last value #662

Closed

PhysSong mentioned this pull request Sep 10, 2017

Automations continue after the end of their TCOs. #3800

Closed

lukas-w mentioned this pull request Oct 1, 2017

Xpressivebackport #3845

Closed

tresf mentioned this pull request Oct 12, 2017

Cant remove dial-memory pos #3877

Closed

PhysSong mentioned this pull request Nov 26, 2017

Handle shrank automation patterns correctly #4012

Merged

musikBear mentioned this pull request Jun 24, 2018

Improve logic for controllers in automation-locked state #4446

Open



		void processNextBuffer();
		void processAutomations(const TrackList& tracks, MidiTime timeStart, fpp_t frames, int tcoNum);

Uh oh!

Conversation

lukas-w commented Feb 23, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tresf commented Feb 23, 2017

Uh oh!

zonkmachine commented Mar 7, 2017

Uh oh!

lukas-w commented Mar 7, 2017

Uh oh!

lukas-w commented Mar 7, 2017

Uh oh!

lukas-w commented Mar 9, 2017

Uh oh!

Umcaruje commented Mar 9, 2017

Uh oh!

tresf commented Mar 9, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

musikBear commented Mar 10, 2017

Uh oh!

musikBear commented Mar 10, 2017

Uh oh!

jasp00 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lukas-w Mar 18, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lukas-w commented Mar 21, 2017

Uh oh!

jasp00 commented Mar 21, 2017

Uh oh!

lukas-w commented Mar 22, 2017

Uh oh!

lukas-w commented Mar 22, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jasp00 commented Mar 23, 2017

Uh oh!

lukas-w commented Mar 23, 2017

Uh oh!

tresf commented Mar 23, 2017

Uh oh!

lukas-w commented Mar 23, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tresf commented Mar 23, 2017

Uh oh!

lukas-w commented Mar 26, 2017

Uh oh!

jasp00 commented Mar 28, 2017

Uh oh!

lukas-w commented Mar 28, 2017

Uh oh!

jasp00 commented Mar 28, 2017

Uh oh!

Reviewers

Assignees

Labels

lukas-w commented Feb 23, 2017 •

edited

Loading

tresf commented Mar 9, 2017 •

edited

Loading

lukas-w Mar 18, 2017 •

edited

Loading

lukas-w commented Mar 22, 2017 •

edited

Loading

lukas-w commented Mar 23, 2017 •

edited

Loading