Bug 401757 - Complex weird behavior with layers and undo history
Summary: Complex weird behavior with layers and undo history
Status: RESOLVED DUPLICATE of bug 403048
Alias: None
Product: krita
Classification: Applications
Component: General (show other bugs)
Version: git master (please specify the git hash!)
Platform: Other Linux
: NOR normal
Target Milestone: ---
Assignee: Krita Bugs
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2018-12-05 01:31 UTC by Storm Engineer
Modified: 2019-01-15 13:53 UTC (History)
2 users (show)

See Also:
Latest Commit:
Version Fixed In:
Sentry Crash Report:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Storm Engineer 2018-12-05 01:31:51 UTC
Sometimes the following behavior happens:

Painting on or changing visibility of a layer doesn't show anymore. Selecting another layer makes the changes appear while creating a rogue "Move" entry in undo history. Undoing stops working. Trying to undo steps and then paint or change visibility of a layer may either result in the changes added after the last undo step instead of the selected one and the creation of another false "Move" undo entry, or it may force the undoing to take effect and apply the action after the selected undo step as it would normally be expected - when this happens the bug does away for the time being.

One time it happened while editing a local selection. This caused the selection to be overwritten, which reflected on its thumbnail, but selecting it still showed the decoration for the state before overwritten. After saving the document and restarting Krita the selection was indeed overwritten just like them thumbnail showed earlier.

The triggering of and behavior of this bug seems inconsistent, so far I was unable to identify any patterns or reliable ways to reproduce it, but it happened several times in the past week or so.

Saving the document and restarting Krita clears the issue but data loss may happen if we were unable to force undoing to take effect before saving.


STEPS TO REPRODUCE
Unknown - but probably related to changing visibility of and/or moving layers.


Krita
  Version: 4.2.0-pre-alpha (git 05baa5b)

Qt
  Version (compiled): 5.11.2
  Version (loaded): 5.11.2

OS Information
  Build ABI: x86_64-little_endian-lp64
  Build CPU: x86_64
  CPU: x86_64
  Kernel Type: linux
  Kernel Version: 4.19.4-arch1-1-ARCH
  Pretty Productname: Arch Linux
  Product Type: arch
  Product Version: unknown

OpenGL Info 
  Vendor:  NVIDIA Corporation 
  Renderer:  "GeForce GTX 750 Ti/PCIe/SSE2" 
  Version:  "4.6.0 NVIDIA 415.18" 
  Shading language:  4.60 NVIDIA 
  Requested format:  QSurfaceFormat(version 3.0, options QFlags<QSurfaceFormat::FormatOption>(DeprecatedFunctions), depthBufferSize 24, redBufferSize -1, greenBufferSize -1, blueBufferSize -1, alphaBufferSize -1, stencilBufferSize 8, samples -1, swapBehavior QSurfaceFormat::SwapBehavior(DoubleBuffer), swapInterval 0, colorSpace QSurfaceFormat::ColorSpace(DefaultColorSpace), profile  QSurfaceFormat::OpenGLContextProfile(CompatibilityProfile)) 
  Current format:    QSurfaceFormat(version 4.6, options QFlags<QSurfaceFormat::FormatOption>(DeprecatedFunctions), depthBufferSize 24, redBufferSize 8, greenBufferSize 8, blueBufferSize 8, alphaBufferSize 0, stencilBufferSize 8, samples -1, swapBehavior QSurfaceFormat::SwapBehavior(DoubleBuffer), swapInterval 0, colorSpace QSurfaceFormat::ColorSpace(DefaultColorSpace), profile  QSurfaceFormat::OpenGLContextProfile(CompatibilityProfile)) 
     Version: 4.6
     Supports deprecated functions true 
     is OpenGL ES: false
Comment 1 Storm Engineer 2019-01-04 16:54:05 UTC
Still experiencing this. It also interferes with saving: the document will give a warning that an operation is still in progress, and when I click save regardless, it won't actually save.

I have just spent 5 minutes doing all kinds of random things hoping that something makes my document savable.

I know this report is not very helpful but I don't know how could I describe it any better.
Comment 2 Storm Engineer 2019-01-05 11:26:24 UTC
This keeps occurring but I still didn't find a consistent way of reproducing.

But my best guess is that it's either triggered by using selections or by toggling layer visibility, or maybe the combination of the two.

I wonder if anyone else experienced this?
Comment 3 Rebecca Breu 2019-01-05 12:18:29 UTC
Yes, get this, too. There was a bit of discussion here, though off-topic for that ticket:

https://bugs.kde.org/show_bug.cgi?id=396509#c12
Comment 4 Storm Engineer 2019-01-05 16:43:11 UTC
I have finally found a consistent way to reproduce:

Open existing file or make a new, just have two layers and at least one with content.

1; Move layer with Move tool
2; Switch back to Brush tool
3; Select the other layer and try to draw.

What happens: The canvas won't update. Toggling layer visibility won't update. Undo won't work. You can't save, add new layer, or perform seve4ral other operations (Popup: Waiting for current operation).

4; Select another layer

What happens: Canvas and layers suddenly update, and a bogus "Move" entry is added to history. This keeps happening every time another layer is selected.


It can temporarily be fixed by doing this:

1; Toggle the visibility of a layer that is currently invisible and NOT selected (it wont't work)
2; Select the layer you toggled

What happens: Canvas and layer updates, you can draw, save, undo, add layer etc. again but only until you select another layer, which will then trigger the bug again.
Comment 5 Storm Engineer 2019-01-05 18:49:41 UTC
First bad commit (via git bisect):

commit 5f91230f074be53379ba1461c6a1560d1f8aa1eb
Author: Boudewijn Rempt <boud@valdyas.org>
Date:   Wed Nov 28 12:44:43 2018 +0100

    add a createActions method to KoToolFactoryBase
Comment 6 Rebecca Breu 2019-01-05 19:10:33 UTC
I can reproduce this with various recent nightly builds. There's probably other combinations of selecting/layer actions leading to this, but the steps above work every time for me.

Krita
  Version: 4.2.0-pre-alpha (git 52afd0c)

Qt
  Version (compiled): 5.10.0
  Version (loaded): 5.10.0

OS Information
  Build ABI: x86_64-little_endian-lp64
  Build CPU: x86_64
  CPU: x86_64
  Kernel Type: linux
  Kernel Version: 4.9.0-8-amd64
  Pretty Productname: Debian GNU/Linux 9 (stretch)
  Product Type: debian
  Product Version: 9

OpenGL Info 
  Vendor:  Intel Open Source Technology Center 
  Renderer:  "Mesa DRI Intel(R) Kabylake GT2 " 
  Version:  "3.0 Mesa 13.0.6" 
  Shading language:  1.30 
  Requested format:  QSurfaceFormat(version 3.0, options QFlags<QSurfaceFormat::FormatOption>(DeprecatedFunctions), depthBufferSize 24, redBufferSize -1, greenBufferSize -1, blueBufferSize -1, alphaBufferSize -1, stencilBufferSize 8, samples -1, swapBehavior QSurfaceFormat::SwapBehavior(DoubleBuffer), swapInterval 0, colorSpace QSurfaceFormat::ColorSpace(DefaultColorSpace), profile  QSurfaceFormat::OpenGLContextProfile(CompatibilityProfile)) 
  Current format:    QSurfaceFormat(version 3.0, options QFlags<QSurfaceFormat::FormatOption>(DeprecatedFunctions), depthBufferSize 24, redBufferSize 8, greenBufferSize 8, blueBufferSize 8, alphaBufferSize 0, stencilBufferSize 8, samples -1, swapBehavior QSurfaceFormat::SwapBehavior(DoubleBuffer), swapInterval 0, colorSpace QSurfaceFormat::ColorSpace(DefaultColorSpace), profile  QSurfaceFormat::OpenGLContextProfile(NoProfile)) 
     Version: 3.0
     Supports deprecated functions true 
     is OpenGL ES: false
Comment 7 Rebecca Breu 2019-01-05 21:16:06 UTC
I wonder if this is due to the same commit: Since about the same time I've had frequent segfaults when confirming moves or transformations. I haven't found a reliable set of steps to reproduce, though, but once this issue is fixed I can try again and see if I still get them.
Comment 8 Rebecca Breu 2019-01-14 20:37:54 UTC
This commit might also be causing this bug:

https://bugs.kde.org/show_bug.cgi?id=403048
Comment 9 Halla Rempt 2019-01-15 11:49:01 UTC

*** This bug has been marked as a duplicate of bug 403048 ***
Comment 10 Halla Rempt 2019-01-15 13:53:04 UTC
It seems that only using the move tool could create this weird state. I've fixed that in https://commits.kde.org/krita/0d2f1d16e187a2110e2bf4b21e0d9c23b0871489