Control links are lost when duplicating automation clips #3781

unfa · 2017-08-27T16:50:03Z

When using Copy/Paste or Ctrl+Drag to duplicate an automation clip, the link to a control is lost.

This makes it very tedious to duplicate automation clips, as one needs to restore the links manually for each clip.

LMMS 1.2 RC3 Linux.

tresf · 2017-08-27T16:51:41Z

+1, hit this all the time. This must be an order of operations problem because once you Ctrl + drag the control back it fixes it but the description copies just fine.

karmux · 2017-08-27T21:19:41Z

This doesn't always happen with me but from time to time. Right now I'm unable to reproduce.
Also related bug is that pasted automation clip will have duplicate connections when target automation clip already has same connection.

musikBear · 2017-08-28T15:25:06Z

This i reported earlier, and it is a renewed fail.
In 1.190 This worked
Automation could be copied by

ctrl-drag
using context-copy
copy-tool
This was iow (re)introduced between 1.1.90 and RC2
The link was Copies of shapes in automation are wrong #3359

musikBear · 2017-09-23T21:30:55Z

Of cause you already knows this, but in case you do not..
The only thing that is missing for the connection to re-established, is the objectID
Here is the xml for only a automation TCO

<automationpattern tens="1" pos="0" prog="2" len="768" mute="0" name="TripleOscillator>Volume">
          <time pos="0" value="100"/>
          <time pos="168" value="165.7"/>
          <time pos="336" value="88.3"/>
          <time pos="456" value="11.7"/>
          <time pos="576" value="115.3"/>
          <object id="31716"/>
        </automationpattern>
        <automationpattern tens="1" pos="1536" prog="2" len="768" mute="0" name="TripleOscillator>Volume">
          <time pos="0" value="100"/>
          <time pos="168" value="165.7"/>
          <time pos="336" value="88.3"/>
          <time pos="456" value="11.7"/>
          <time pos="576" value="115.3"/>
		  
        </automationpattern>

If i simply copy the objectID into the second block, like this

        <automationpattern tens="1" pos="0" prog="2" len="768" mute="0" name="TripleOscillator>Volume">
          <time pos="0" value="100"/>
          <time pos="168" value="165.7"/>
          <time pos="336" value="88.3"/>
          <time pos="456" value="11.7"/>
          <time pos="576" value="115.3"/>
          <object id="31716"/>
        </automationpattern>
        <automationpattern tens="1" pos="1536" prog="2" len="768" mute="0" name="TripleOscillator>Volume">
          <time pos="0" value="100"/>
          <time pos="168" value="165.7"/>
          <time pos="336" value="88.3"/>
          <time pos="456" value="11.7"/>
          <time pos="576" value="115.3"/>
		  <object id="31716"/>
        </automationpattern>

The TCO get the curve shape back, and works on the automated control
<object id= /> is for some reason, not copied when a automation TCO is ctrl+drag- copied

PhysSong · 2017-11-30T02:17:10Z

This is a regression in 1.2, caused by #2875. When I reverted f97b431, the bug was gone.
However, We shouldn't simply revert that commit because it will reintroduce journaling ID collisions. I'm looking for good ideas...

NETMANSKY · 2018-06-07T21:42:37Z

Well is it fixed?

PhysSong · 2018-06-08T01:15:49Z

Not yet, but it will be likely to be fixed in RC7.

PhysSong · 2018-07-24T06:57:39Z

Here are some ideas:

(temporary) Revert Split journal ID range between new and loaded elements #2875. Fixing a major bug can be preferable to fixing a potential issue.
(temporary) Use idToSave only if saving projects. The issue described in Split journal ID range between new and loaded elements #2875 remains.
Decouple automation from journaling. Maybe a master thing, but it will be much better if implemented correctly.

On a side note, #2875 doesn't fix the collision perfectly. For example, you can loading data/presets/Kicker/SnareMarch.xpf or data/presets/TripleOscillator/DetunedGhost.xpf from sidebar.

zonkmachine · 2018-07-24T13:44:37Z

Something like

(temporary) Revert Split journal ID range between new and loaded elements #2875. Fixing a major bug can be preferable to fixing a potential issue.

... for rc7 and

Decouple automation from journaling. Maybe a master thing, but it will be much better if implemented correctly.

... for the final 1.2 release?

musikBear · 2018-07-24T21:08:55Z

May i ask why the <object id="...."/>is lost from the datastructure?
Is that understood, and not repairable?
-why not reuse of a string concat ... idk.. :| should propl shut up.. 🙊

PhysSong · 2018-07-25T05:33:53Z

May i ask why the <object id="...."/> is lost from the datastructure?

@musikBear After #2875, new objects and saved objects use different ranges of ID. When saving, new objects' IDs are converted by idToSave function.
When you copy an automation pattern, LMMS saves the pattern's data and reloads it. While the process, the ID conversion occurs and LMMS will try to find objects by wrong IDs. So the connection can't be re-established and the missing block is the consequence of it.

musikBear · 2018-07-25T17:11:42Z

@PhysSong

" LMMS will try to find objects by wrong IDs

Oh.. i understand. Good explanation 👍

zonkmachine · 2018-07-31T03:14:02Z

(temporary) Revert Split journal ID range between new and loaded elements #2875. Fixing a major bug can be preferable to fixing a potential issue.

Yeah, I think this is a good idea. Revert #2875 and remove this issue from RC7. Unless you have a better fix in the works off course.

PhysSong · 2018-07-31T06:41:22Z

Unless you have a better fix in the works off course.

I have a fix, but I think it needs some changes. I'll remove this from RC7 without doing anything, but keep this in 1.2.0 milestone.

PhysSong · 2018-10-10T13:29:15Z

How about hacking AutomationPattern::resolveAllIDs? I think we can have a simple fallback logic which tries adding EO_ID_MSB to the ID. It fixes the symptom, but not the cause.
Note that some structural changes are needed to fix the underlying bug completely.

diff --git a/src/core/AutomationPattern.cpp b/src/core/AutomationPattern.cpp
index 25da6defb..c14b83104 100644
--- a/src/core/AutomationPattern.cpp
+++ b/src/core/AutomationPattern.cpp
@@ -787,6 +787,14 @@ void AutomationPattern::resolveAllIDs()
 						{
 							a->addObject( dynamic_cast<AutomatableModel *>( o ), false );
 						}
+						else
+						{
+							o = Engine::projectJournal()->journallingObject( *k + ( 1 << 23 ) );
+							if( o && dynamic_cast<AutomatableModel *>( o ) )
+							{
+								a->addObject( dynamic_cast<AutomatableModel *>( o ), false );
+							}
+						}
 					}
 					a->m_idsToResolve.clear();
 					a->dataChanged();

DomClark · 2018-12-08T18:02:55Z

It fixes the symptom, but not the cause.
Note that some structural changes are needed to fix the underlying bug completely.

I'm not too familiar with the journalling code, so I can't comment on whether it's a good fix or not, but if it works, you're happy with it, and we'd like to minimise the number of structural changes in stable-1.2, then I'd say go for it. If I'm following things correctly, this is one of the main blockers for 1.2, so fixing the just the symptoms would be better than no fix at all. Structural changes can always be done in master.

PhysSong · 2018-12-09T07:21:09Z

and we'd like to minimise the number of structural changes in stable-1.2

Agreed. Since I don't have a better fix for stable-1.2, I've decided to use the fix above.

musikBear · 2018-12-11T14:04:16Z

The 'lost connection' is unfortunately also a fact for on/off controllers, and not just for copies, but also after reopen of a saved project.
Recreate :
Automate any FX-chain On/Off
Replay - FX-chain follows automation
Save project
Reload project
Replay -FX-chain will not follow automation
( Redrag of FX-chain On/Off into automation, Restores functionality -But thats rather obvious..

PhysSong · 2018-12-11T15:05:43Z

@musikBear I think you're talking about #4632 (comment) .

musikBear · 2018-12-11T19:28:59Z

@musikBear I think you're talking about #4632 (comment) .
:: Whether to enable effect chains

@PhysSong
Oh yes -that is the one
I am next to sure, earlier versions has had on/off automation capabilities, for effect-chain, and its not that it cant be automated, because it can, but reloaded projects have lost this automation, even though the ID for controller is not lost:
´´´´
Its just not reacting (?)

PhysSong · 2018-12-12T12:38:26Z

I didn't explain that in #4632: IDs are saved correctly in automation patterns, but not in connected models(the green LED). If any of them doesn't contain its ID, the automation will be broken.

PhysSong mentioned this issue Nov 23, 2017

Gltich causing Automation Tracks to forget their connections #4001

Closed

PhysSong added this to the 1.2.0 milestone Nov 30, 2017

zonkmachine added the bug label Dec 11, 2017

tresf mentioned this issue Feb 28, 2018

Copies of shapes in automation are wrong #3359

Closed

PhysSong mentioned this issue May 27, 2018

Fix journaling system bugs #4377

Closed

PhysSong added this to To do in Release LMMS 1.2.0-RC7 Jun 3, 2018

DomClark mentioned this issue Jun 7, 2018

Automatization copy loses a connection #4409

Closed

PhysSong removed this from To do in Release LMMS 1.2.0-RC7 Jul 31, 2018

PhysSong added this to To do in Release LMMS 1.2.0-RC8 Aug 7, 2018

PhysSong mentioned this issue Sep 27, 2018

LMMS 1.2.0 rc7 some little things that are overlooked #4626

Closed

tresf mentioned this issue Sep 29, 2018

Automating "mute" is broken on project load #4632

Closed

PhysSong mentioned this issue Oct 27, 2018

Automations lose it's connection and the shape if you copy them in the Song editor #4681

Closed

PhysSong mentioned this issue Dec 9, 2018

Fix lost control links in copied automation patterns #4723

Merged

zonkmachine moved this from To do to In progress in Release LMMS 1.2.0-RC8 Dec 9, 2018

PhysSong closed this as completed in #4723 Dec 22, 2018

Release LMMS 1.2.0-RC8 automation moved this from In progress to Done Dec 22, 2018

PhysSong mentioned this issue Feb 21, 2019

LADSPA automation disconnects on export #4781

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Control links are lost when duplicating automation clips #3781

Control links are lost when duplicating automation clips #3781

unfa commented Aug 27, 2017

tresf commented Aug 27, 2017 •

edited

Loading

karmux commented Aug 27, 2017

musikBear commented Aug 28, 2017 •

edited

Loading

musikBear commented Sep 23, 2017

PhysSong commented Nov 30, 2017 •

edited

Loading

NETMANSKY commented Jun 7, 2018

PhysSong commented Jun 8, 2018

PhysSong commented Jul 24, 2018

zonkmachine commented Jul 24, 2018

musikBear commented Jul 24, 2018

PhysSong commented Jul 25, 2018

musikBear commented Jul 25, 2018

zonkmachine commented Jul 31, 2018

PhysSong commented Jul 31, 2018

PhysSong commented Oct 10, 2018

DomClark commented Dec 8, 2018

PhysSong commented Dec 9, 2018

musikBear commented Dec 11, 2018

PhysSong commented Dec 11, 2018

musikBear commented Dec 11, 2018

PhysSong commented Dec 12, 2018

Control links are lost when duplicating automation clips #3781

Control links are lost when duplicating automation clips #3781

Comments

unfa commented Aug 27, 2017

tresf commented Aug 27, 2017 • edited Loading

karmux commented Aug 27, 2017

musikBear commented Aug 28, 2017 • edited Loading

musikBear commented Sep 23, 2017

PhysSong commented Nov 30, 2017 • edited Loading

NETMANSKY commented Jun 7, 2018

PhysSong commented Jun 8, 2018

PhysSong commented Jul 24, 2018

zonkmachine commented Jul 24, 2018

musikBear commented Jul 24, 2018

PhysSong commented Jul 25, 2018

musikBear commented Jul 25, 2018

zonkmachine commented Jul 31, 2018

PhysSong commented Jul 31, 2018

PhysSong commented Oct 10, 2018

DomClark commented Dec 8, 2018

PhysSong commented Dec 9, 2018

musikBear commented Dec 11, 2018

PhysSong commented Dec 11, 2018

musikBear commented Dec 11, 2018

PhysSong commented Dec 12, 2018

tresf commented Aug 27, 2017 •

edited

Loading

musikBear commented Aug 28, 2017 •

edited

Loading

PhysSong commented Nov 30, 2017 •

edited

Loading