action editor in dope sheet cant be deleted after created new one #45535

Closed
opened 2015-07-23 05:45:53 +02:00 by yellow foxz · 20 comments

System Information
Operating system and graphics card

Blender Version
2.75a hashc27589e

Short description of error
action dope sheet cant be deleted after created new one

Exact steps for others to reproduce the error
1 open blender and add object in the scene
2 change to dope sheet editor
3 select action editor mode in dope sheet
4 click new
5 click "+" to create new one, must click multiple times
6 delete the previous one or rest of them
this file>> bug Action dope sheet.rar

**System Information** Operating system and graphics card **Blender Version** 2.75a hashc27589e **Short description of error** action dope sheet cant be deleted after created new one **Exact steps for others to reproduce the error** 1 open blender and add object in the scene 2 change to dope sheet editor 3 select action editor mode in dope sheet 4 click new 5 click "+" to create new one, must click multiple times 6 delete the previous one or rest of them this file>> [bug Action dope sheet.rar](https://archive.blender.org/developer/F211774/bug_Action_dope_sheet.rar)
Author

Changed status to: 'Open'

Changed status to: 'Open'
Author

Added subscriber: @yellowfoxz

Added subscriber: @yellowfoxz

#45554 was marked as duplicate of this issue

#45554 was marked as duplicate of this issue
Member

Added subscriber: @JulianEisel

Added subscriber: @JulianEisel
Member

Seems like this is the same issue as reported in #45554 - merging

Seems like this is the same issue as reported in #45554 - merging
Member

Added subscriber: @motorsep

Added subscriber: @motorsep

Can developers confirm the bug or is it going to be one of those who no one wants to touch with a long stick? (like it was with losing work if user wasn't manually assigned a few version back)

Can developers confirm the bug or is it going to be one of those who no one wants to touch with a long stick? (like it was with losing work if user wasn't manually assigned a few version back)

New development about this bug. Apparently, if Action is duplicated, but user is never assigned to it, it can be deleted by deleting it from NLA Editor (it's still backward, as there is no reason not to be able to remove Actions from Action Editor). Here is a video showing it:

https://www.youtube.com/watch?v=Z3H_RDnDf-g

However, if I click on F next to the Action I will need to delete, it's a done deal - it's there to stay no matter what I do.

New development about this bug. Apparently, if Action is duplicated, but user is never assigned to it, it can be deleted by deleting it from NLA Editor (it's still backward, as there is no reason not to be able to remove **Actions** from **Action Editor**). Here is a video showing it: https://www.youtube.com/watch?v=Z3H_RDnDf-g However, if I click on F next to the Action I will need to delete, it's a done deal - it's there to stay no matter what I do.

It gets even weirder - the user who posted video showing how Actions that never got F pressed can be removed, can remove Actions even after having users on them. I followed his instructions to the tee, but when I save, users get assigned to the actions automatically:

https://www.youtube.com/watch?v=mtGTbqgnNJs

It gets even weirder - the user who posted video showing how Actions that never got F pressed can be removed, can remove Actions even after having users on them. I followed his instructions to the tee, but when I save, users get assigned to the actions automatically: https://www.youtube.com/watch?v=mtGTbqgnNJs
Member

Added subscriber: @JoshuaLeung

Added subscriber: @JoshuaLeung
Member

Can developers confirm the bug or is it going to be one of those who no one wants to touch with a long stick? (like it was with losing work if user wasn't manually assigned a few version back)

Such a statement is complete rubbish and I hope you know that. If an issue isn't fixed within a reasonable amount of time than because it's not easy to fix it. The example you mention, possible loss of actions, was broken by design - a design that is deeply implemented into Blender's data structure. AFAIK it's still not fixed (which would require redesigning Blender's datablock system), but there were tools implemented that help prevent data loss.

Anyway... this is not a forum -> back to topic...

@JoshuaLeung, could you have a look? Not really an area I'm familiar with ;)

> Can developers confirm the bug or is it going to be one of those who no one wants to touch with a long stick? (like it was with losing work if user wasn't manually assigned a few version back) Such a statement is complete rubbish and I hope you know that. If an issue isn't fixed within a reasonable amount of time than because it's not easy to fix it. The example you mention, possible loss of actions, was broken by design - a design that is deeply implemented into Blender's data structure. AFAIK it's still not fixed (which would require redesigning Blender's datablock system), but there were tools implemented that help prevent data loss. Anyway... this is not a forum -> back to topic... @JoshuaLeung, could you have a look? Not really an area I'm familiar with ;)

In #45535#323859, @JulianEisel wrote:
Such a statement is complete rubbish and I hope you know that. If an issue isn't fixed within a reasonable amount of time than because it's not easy to fix it. The example you mention, possible loss of actions, was broken by design - a design that is deeply implemented into Blender's data structure. AFAIK it's still not fixed (which would require redesigning Blender's datablock system), but there were tools implemented that help prevent data loss.

My bad, sorry.

I am wondering if there are user settings that might interfere with functionality (and thus some users don't have an issue and some do). I also wonder why not to just leave Shift +- [x] be as it worked for everyone as a good solution.

> In #45535#323859, @JulianEisel wrote: > Such a statement is complete rubbish and I hope you know that. If an issue isn't fixed within a reasonable amount of time than because it's not easy to fix it. The example you mention, possible loss of actions, was broken by design - a design that is deeply implemented into Blender's data structure. AFAIK it's still not fixed (which would require redesigning Blender's datablock system), but there were tools implemented that help prevent data loss. My bad, sorry. I am wondering if there are user settings that might interfere with functionality (and thus some users don't have an issue and some do). I also wonder why not to just leave Shift +- [x] be as it worked for everyone as a good solution.

Added subscriber: @zeauro

Added subscriber: @zeauro

It looks like duplication was thought to be done only once.
If you get rid of action active in Action Editor (by pushing it, stashing it, deleting it or replacing it by another), there is no stashing in NLA at new action creation.

It looks like duplication was thought to be done only once. If you get rid of action active in Action Editor (by pushing it, stashing it, deleting it or replacing it by another), there is no stashing in NLA at new action creation.
Member

There are several things going on here, so I'll go through them one by one to avoid confusing things further.

yellowfox's file

From the sound of things, it looks like this is just a case of not realising that we now hang on to all actions in the NLA.

Z3H_RDnDf-g

It looks like everything is behaving exactly as designed. The action stashing behaviours were put in place specifically to address complaints by various users that they were finding it far too easy to accidentally lose actions. In particular, the biggest complaints related to:

  1. Losing an action when starting work on the next one
    1a) The user firstly closes/unlinks the existing action, then presses "Add New" or simply inserts some new keyframes --> A "clean" start
    1b) The user simply presses the "Add New" with the old action still active --> Easier/Lazier way of 1a OR a simple way to preserve an earlier version of the action
 In both of these cases, there are no problems if *something* causes the usercount of the old action to remain non-zero once it has been displaced to make way for the newly created action. A Fake User forcibly gives the action a user (effectively pinning it on a global/file level), whereas Stashing in the NLA means that the Object/Mesh/Material/etc. still has some link to the actions which can be used to affect it (benefits of doing this are that renaming bones or doing other rig changes can get applied to the actions as well).
From many of the threads discussing these things, one of the notions that came up repeatedly was that when you create actions, those are tied to the objects/materials/etc. that you created those actions for. That is, Blender should know that Actions 1, 2, 3 are for Object A, while Actions 4, 5, 6 are for Object B. The action stashing workflow seemed to be a good way to make Blender's behaviour match up with semantic expectations. 
  1. Losing an action when switching between different actions while working on a new one. (Or alternatively, working on multiple actions at once).
The user already has some actions created (A1, A2, A3). They then create a new action (A4), and lay down 1+ keyframes, realise that they need to check on a previous action, then use the Browse button to switch to a previous action (A2/A3).   After checking that previous action, they jump back to the new action they were creating (A4) and carry on.  However, they may be interrupted at any stage of this process, and may end up saving the file with A2/A3 active, but without having done anything to preserve A4 yet.
  1. "Closing" an action after checking on it
 As part of 2), the user may get distracted after A2, and want to start creating a new action (A5) from scratch (instead of using A2 as a base, or continuing to work on A4 for the time being). So, instead, they "unlink/close" A2.
In this case, we clearly don't want A2 to suddenly have nothing referring to it anymore. 

Now, were some of these methods overly heavy-handed? Perhaps...

There is an inherent conflict/tradeoff here between protecting user data and making it hard/cumbersome to get rid of data you don't want anymore. This arises because the computer ultimately cannot possibly know what your intentions are.

  • With great power comes great responsibility: At one end of the spectrum, you put the user in full control. That is, the computer does as you tell it to - it will save when you tell it to save, and it will not attempt to retain things that you've told it you currently don't have a use for.
  • On the other end of the spectrum, if you apply too many safely precautions against the user shooting themselves in the foot, it becomes a cumbersome process to get anything done, as now users have to fight against the protections designed to keep them from doing anything unintentionally. (Note: Even something as passive as a warning prompt in one context can be a massive PITA for another)

Given the feedback that was trickling in - i.e. "data shouldn't ever be lost", "users should tell the software when they want something deleted" - the available evidence pointed to a need to be conservative/cautious here and lean towards saving things more often.

mtGTbqgnNJs

OK, so if I understand correctly, the problem here is that the first three actions (named "0", "0.0001", "0.0002") which are shown as initially having 0 users before saving, are shown as having a Fake User after saving?

What is the exact sequence of steps taken to do reproduce this from a clean file? This shouldn't be happening, but I haven't found a way to reproduce this yet.

Shift X

OK, I guess it's reasonable to make some tweaks here so that if shift-clicking on the X will do the following:

  • Unlink the action from the active-action slot (i.e. the standard "delete" behaviour)
  • Remove fake user
  • Remove the NLA strip used to stash that action - It will ONLY remove the strip if it's used for stashing (i.e. not if it is actually used in the NLA normally).

Is this what you'd expect?

There are several things going on here, so I'll go through them one by one to avoid confusing things further. # yellowfox's file From the sound of things, it looks like this is just a case of not realising that we now hang on to all actions in the NLA. # Z3H_RDnDf-g It looks like everything is behaving exactly as designed. The action stashing behaviours were put in place specifically to address complaints by various users that they were finding it far too easy to accidentally lose actions. In particular, the biggest complaints related to: 1) Losing an action when starting work on the next one 1a) The user firstly closes/unlinks the existing action, then presses "Add New" or simply inserts some new keyframes --> A "clean" start 1b) The user simply presses the "Add New" with the old action still active --> Easier/Lazier way of 1a OR a simple way to preserve an earlier version of the action ``` In both of these cases, there are no problems if *something* causes the usercount of the old action to remain non-zero once it has been displaced to make way for the newly created action. A Fake User forcibly gives the action a user (effectively pinning it on a global/file level), whereas Stashing in the NLA means that the Object/Mesh/Material/etc. still has some link to the actions which can be used to affect it (benefits of doing this are that renaming bones or doing other rig changes can get applied to the actions as well). ``` ``` From many of the threads discussing these things, one of the notions that came up repeatedly was that when you create actions, those are tied to the objects/materials/etc. that you created those actions for. That is, Blender should know that Actions 1, 2, 3 are for Object A, while Actions 4, 5, 6 are for Object B. The action stashing workflow seemed to be a good way to make Blender's behaviour match up with semantic expectations. ``` 2) Losing an action when switching between different actions while working on a new one. (Or alternatively, working on multiple actions at once). ``` The user already has some actions created (A1, A2, A3). They then create a new action (A4), and lay down 1+ keyframes, realise that they need to check on a previous action, then use the Browse button to switch to a previous action (A2/A3). After checking that previous action, they jump back to the new action they were creating (A4) and carry on. However, they may be interrupted at any stage of this process, and may end up saving the file with A2/A3 active, but without having done anything to preserve A4 yet. ``` 3) "Closing" an action after checking on it ``` As part of 2), the user may get distracted after A2, and want to start creating a new action (A5) from scratch (instead of using A2 as a base, or continuing to work on A4 for the time being). So, instead, they "unlink/close" A2. ``` ``` In this case, we clearly don't want A2 to suddenly have nothing referring to it anymore. ``` Now, were some of these methods overly heavy-handed? Perhaps... There is an inherent conflict/tradeoff here between protecting user data and making it hard/cumbersome to get rid of data you don't want anymore. This arises because the computer ultimately cannot possibly know what your intentions are. * With great power comes great responsibility: At one end of the spectrum, you put the user in full control. That is, the computer does as you tell it to - it will save when you tell it to save, and it will not attempt to retain things that you've told it you currently don't have a use for. * On the other end of the spectrum, if you apply too many safely precautions against the user shooting themselves in the foot, it becomes a cumbersome process to get anything done, as now users have to fight against the protections designed to keep them from doing anything unintentionally. (Note: Even something as passive as a warning prompt in one context can be a massive PITA for another) Given the feedback that was trickling in - i.e. "data shouldn't ever be lost", "users should tell the software when they want something deleted" - the available evidence pointed to a need to be conservative/cautious here and lean towards saving things more often. # mtGTbqgnNJs OK, so if I understand correctly, the problem here is that the first three actions (named "0", "0.0001", "0.0002") which are shown as initially having 0 users before saving, are shown as having a Fake User after saving? What is the exact sequence of steps taken to do reproduce this from a clean file? This shouldn't be happening, but I haven't found a way to reproduce this yet. # Shift X OK, I guess it's reasonable to make some tweaks here so that if shift-clicking on the X will do the following: * Unlink the action from the active-action slot (i.e. the standard "delete" behaviour) * Remove fake user * Remove the NLA strip used to stash that action - It will ONLY remove the strip if it's used for stashing (i.e. not if it is actually used in the NLA normally). Is this what you'd expect?

So I tried making simple scene from scratch to reproduce the bug, and I easily reproduced it.

Blend file: action_bug_test.blend

All I did was:

  1. Created a mesh
  2. Created Armature
  3. Named root bone "origin", made a chain of 3 bones, parent the chain to "origin"
  4. Parented mesh to Armature with autoweights
  5. Split view and open Action Editor
  6. Select all bones in Pose mode, create a new Action, name it "init" key LocRot on frame 1, and then on frame 2
  7. Click F button on "init" Action
  8. Click + to create a new Action, rename it, click F
  9. Make a simple animation
  10. Click + to create new Action, then F
  11. Do #10 two more times
  12. Save .blend file
  13. Quit Blender and restart it
  14. Open blend file from #12
  15. Click F to remove users on Action created in #10 and #11
  16. Open NLA Editor
  17. Delete the NDA strips for Actions from #10 and #11 (or you can kill the whole thing in NLA)
  18. Save again
  19. Observer unwanted Actions to have users again.

I never had issues with 2.74 - Shift + X button deleted Action, and aster saving/restarting Blender it was gone.

So I tried making simple scene from scratch to reproduce the bug, and I easily reproduced it. Blend file: [action_bug_test.blend](https://archive.blender.org/developer/F218568/action_bug_test.blend) All I did was: 1. Created a mesh 2. Created Armature 3. Named root bone "origin", made a chain of 3 bones, parent the chain to "origin" 4. Parented mesh to Armature with autoweights 5. Split view and open Action Editor 6. Select all bones in Pose mode, create a new Action, name it "init" key LocRot on frame 1, and then on frame 2 7. Click F button on "init" Action 8. Click + to create a new Action, rename it, click F 9. Make a simple animation 10. Click + to create new Action, then F 11. Do #10 two more times 12. Save .blend file 13. Quit Blender and restart it 14. Open blend file from #12 15. Click F to remove users on Action created in #10 and #11 16. Open NLA Editor 17. Delete the NDA strips for Actions from #10 and #11 (or you can kill the whole thing in NLA) 18. Save again 19. Observer unwanted Actions to have users again. I never had issues with 2.74 - Shift + X button deleted Action, and aster saving/restarting Blender it was gone.
Member

Unfortunately, I still can't reproduce any errors here. The actions in question are getting deleted as expected here after following the steps you've listed. Does this still happen in current master/buildbot builds?

I'll keep looking into this though, and will be making the Shift-X improvements proposed above.

Unfortunately, I still can't reproduce any errors here. The actions in question are getting deleted as expected here after following the steps you've listed. Does this still happen in current master/buildbot builds? I'll keep looking into this though, and will be making the Shift-X improvements proposed above.

Added subscriber: @Sergey

Added subscriber: @Sergey

Changed status from 'Open' to: 'Archived'

Changed status from 'Open' to: 'Archived'
Sergey Sharybin self-assigned this 2016-07-29 12:25:54 +02:00

First of all, no feedback on Joshua's request to test with latest builds. Since there are high possibility of the issue was already fixed it's crucial to test newest builds. So archiving for until it is confirmed the issue can be reproduced with latest builds from builder.blender.org.

P.S. Also seems Joshua made all planned modifications in the code now.

First of all, no feedback on Joshua's request to test with latest builds. Since there are high possibility of the issue was already fixed it's crucial to test newest builds. So archiving for until it is confirmed the issue can be reproduced with latest builds from builder.blender.org. P.S. Also seems Joshua made all planned modifications in the code now.
Sign in to join this conversation.
No Label
Interest
Alembic
Interest
Animation & Rigging
Interest
Asset Browser
Interest
Asset Browser Project Overview
Interest
Audio
Interest
Automated Testing
Interest
Blender Asset Bundle
Interest
BlendFile
Interest
Collada
Interest
Compatibility
Interest
Compositing
Interest
Core
Interest
Cycles
Interest
Dependency Graph
Interest
Development Management
Interest
EEVEE
Interest
EEVEE & Viewport
Interest
Freestyle
Interest
Geometry Nodes
Interest
Grease Pencil
Interest
ID Management
Interest
Images & Movies
Interest
Import Export
Interest
Line Art
Interest
Masking
Interest
Metal
Interest
Modeling
Interest
Modifiers
Interest
Motion Tracking
Interest
Nodes & Physics
Interest
OpenGL
Interest
Overlay
Interest
Overrides
Interest
Performance
Interest
Physics
Interest
Pipeline, Assets & IO
Interest
Platforms, Builds & Tests
Interest
Python API
Interest
Render & Cycles
Interest
Render Pipeline
Interest
Sculpt, Paint & Texture
Interest
Text Editor
Interest
Translations
Interest
Triaging
Interest
Undo
Interest
USD
Interest
User Interface
Interest
UV Editing
Interest
VFX & Video
Interest
Video Sequencer
Interest
Virtual Reality
Interest
Vulkan
Interest
Wayland
Interest
Workbench
Interest: X11
Legacy
Blender 2.8 Project
Legacy
Milestone 1: Basic, Local Asset Browser
Legacy
OpenGL Error
Meta
Good First Issue
Meta
Papercut
Meta
Retrospective
Meta
Security
Module
Animation & Rigging
Module
Core
Module
Development Management
Module
EEVEE & Viewport
Module
Grease Pencil
Module
Modeling
Module
Nodes & Physics
Module
Pipeline, Assets & IO
Module
Platforms, Builds & Tests
Module
Python API
Module
Render & Cycles
Module
Sculpt, Paint & Texture
Module
Triaging
Module
User Interface
Module
VFX & Video
Platform
FreeBSD
Platform
Linux
Platform
macOS
Platform
Windows
Priority
High
Priority
Low
Priority
Normal
Priority
Unbreak Now!
Status
Archived
Status
Confirmed
Status
Duplicate
Status
Needs Info from Developers
Status
Needs Information from User
Status
Needs Triage
Status
Resolved
Type
Bug
Type
Design
Type
Known Issue
Type
Patch
Type
Report
Type
To Do
No Milestone
No project
No Assignees
7 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: blender/blender#45535
No description provided.