search

pyvista / pyvista

3D plotting and mesh analysis through a streamlined interface for the Visualization Toolkit (VTK)
https://docs.pyvista.org
MIT License
2.76k stars 509 forks source link

`auto_close=False` not functioning #1621

Open mcgoldba opened 3 years ago

mcgoldba commented 3 years ago

Describe the bug, what's wrong, and what you expect:

I have a number of existing scripts that worked on a Windows machine and an older version of pyVista (I do not know which version as that computer had hard drive issues, and was cleaned).

The problem seems to be that the initial call to p.show(auto_close=False) still closes the plotter, and subsequent calls to p.show(screenshot='image.png') return an error.

To Reproduce

This behavior is reproduced with the code included in the PyVista Plotter() documentation:

import pyvista as pv
pl = pv.Plotter()
_ = pl.add_mesh(pv.Cube())
pl.show(auto_close=False)  
pl.show(screenshot='my_image.png')

which results in the following error after closing the interactive window (by pressing the q button):

python test.py 
/home/mcgoldba/miniconda3/envs/xcfdv/lib/python3.9/site-packages/pyvista-0.31.3-py3.9.egg/pyvista/plotting/plotting.py:4070: UserWarning: `auto_close` ignored: by clicking the exit button, you have destroyed the render window and we have to close it out.
  warnings.warn("`auto_close` ignored: by clicking the exit button, "
Traceback (most recent call last):
  File "/home/mcgoldba/OpenFOAM/mgoldbach-7/run/202000045/sim04/test.py", line 7, in <module>
    pl.show(screenshot='test_img.png')
  File "/home/mcgoldba/miniconda3/envs/xcfdv/lib/python3.9/site-packages/pyvista-0.31.3-py3.9.egg/pyvista/plotting/plotting.py", line 3988, in show
    raise RuntimeError("This plotter has been closed and cannot be shown.")
RuntimeError: This plotter has been closed and cannot be shown.

System Information:

# Get system info
import pyvista as pv
print(pv.Report())
--------------------------------------------------------------------------------
  Date: Mon Aug 30 11:26:55 2021 EDT

                OS : Linux
            CPU(s) : 24
           Machine : x86_64
      Architecture : 64bit
       Environment : Python
        GPU Vendor : NVIDIA Corporation
      GPU Renderer : Quadro M2000/PCIe/SSE2
       GPU Version : 4.5.0 NVIDIA 460.84

  Python 3.9.5 (default, Jun  4 2021, 12:28:51)  [GCC 7.5.0]

           pyvista : 0.31.3
               vtk : 9.0.20210710
             numpy : 1.21.0
           imageio : 2.9.0
           appdirs : 1.4.4
            scooby : 0.5.7
            meshio : 4.4.6
        matplotlib : 3.4.2
          colorcet : 1.0.0
--------------------------------------------------------------------------------
akaszynski commented 3 years ago

I've noticed this on my end as well and I'm looking into it.

mcgoldba commented 3 years ago

It seems that this is still an issue with the newest release of pyvista (0.32.1). However, I have noticed that the issue is with the following lines of code in plotting.py (or maybe in the call to the IsCurrent() vtk function):

https://github.com/pyvista/pyvista/blob/4b83f5c92068e8e6e8e27dc8b5fd974b56fbc7ee/pyvista/plotting/plotting.py#L4949-L4956

If I comment out these lines, the script above works. Since, as I understand, this check is only performed for the benefit of the Windows operating system, can it rather check for the operating system, and then check for the value of IsCurrent()? Something similar to:

if os.name =='nt' and not self.ren_win.IsCurrent():
    ...

Just for completeness, in my instance, the call to self.ren_win returns a vtkXOpenGLRenderWindow, and the call to IsCurrent() always returns False:

import pyvista as pv

print(pv.Report())

pl = pv.Plotter()
_ = pl.add_mesh(pv.Cube())
print(pl.ren_win)
print(pl.ren_win.IsCurrent())
pl.show(auto_close=False)
print(pl.ren_win.IsCurrent())
pl.show(screenshot='my_image.png')
--------------------------------------------------------------------------------
  Date: Thu Oct 07 15:00:56 2021 EDT

                OS : Linux
            CPU(s) : 24
           Machine : x86_64
      Architecture : 64bit
       Environment : Python
        GPU Vendor : NVIDIA Corporation
      GPU Renderer : Quadro M2000/PCIe/SSE2
       GPU Version : 4.5.0 NVIDIA 460.84

  Python 3.9.5 (default, Jun  4 2021, 12:28:51)  [GCC 7.5.0]

           pyvista : 0.32.1
               vtk : 9.0.20210710
             numpy : 1.21.0
           imageio : 2.9.0
           appdirs : 1.4.4
            scooby : 0.5.7
            meshio : 4.4.6
        matplotlib : 3.4.2
          colorcet : 1.0.0
--------------------------------------------------------------------------------
vtkXOpenGLRenderWindow (0x55641b74d630)
  Debug: Off
  Modified Time: 4443
  Reference Count: 2
  Registered Events: (none)
  Erase: On
  Window Name: Visualization Toolkit - OpenGL
  Position: (0, 0)
  Size: (1024, 768)
  Mapped: 0
  ShowWindow: 1
  UseOffScreenBuffers: 0
  Double Buffered: 1
  DPI: 72
  TileScale: (1, 1)
  TileViewport: (0, 0, 1, 1)
  Borders: On
  Double Buffer: On
  Full Screen: Off
  Renderers:
    Debug: Off
    Modified Time: 3401
    Reference Count: 1
    Registered Events: (none)
    Number Of Items: 2
  Stereo Capable Window Requested: No
  Stereo Render: Off
  Point Smoothing: Off
  Line Smoothing: Off
  Polygon Smoothing: Off
  Abort Render: 0
  Current Cursor: 0
  Desired Update Rate: 0.0001
  In Abort Check: 0
  NeverRendered: 1
  Interactor: 0x55641b74c3a0
  Swap Buffers: On
  Stereo Type: RedBlue
  Number of Layers: 2
  AlphaBitPlanes: On
  UseSRGBColorSpace: Off
  AnaglyphColorSaturation: 0.65
  AnaglyphColorMask: 4 , 3
  MultiSamples: 4
  StencilCapable: False
  ContextId: 0
  Color Map: 0
  Display Id: 0
  Next Window Id: 0
  Window Id: 0

False
False
akaszynski commented 3 years ago

If I comment out these lines, the script above works. Since, as I understand, this check is only performed for the benefit of the Windows operating system, can it rather check for the operating system, and then check for the value of IsCurrent()?

That's a good workaround. I'm finding it annoying locally that auto_close isn't working.

banesullivan commented 3 years ago

I just ran into this as well

beroda commented 1 year ago

@mcgoldba Do you still experiences this issue with the following code ? (taken from documentation here).

import pyvista as pv
pl = pv.Plotter()
_ = pl.add_mesh(pv.Cube())
pl.show(auto_close=False)
pl.show(screenshot='my_image.png')

On my side, I obtain the exact same error as you when you posted this issue. My pv.Report():

--------------------------------------------------------------------------------
  Date: Fri Jun 23 18:28:22 2023 CEST

                OS : Linux
            CPU(s) : 8
           Machine : x86_64
      Architecture : 64bit
               RAM : 15.5 GiB
       Environment : Jupyter
       File system : ext4
        GPU Vendor : Intel
      GPU Renderer : Mesa Intel(R) HD Graphics 630 (KBL GT2)
       GPU Version : 4.6 (Core Profile) Mesa 21.2.6
  MathText Support : False

  Python 3.9.17 (main, Jun  6 2023, 20:11:04)  [GCC 9.4.0]

           pyvista : 0.40.dev0
               vtk : 9.2.6
             numpy : 1.25.0
        matplotlib : 3.7.1
            scooby : 0.7.2
             pooch : v1.7.0
           imageio : 2.31.1
           IPython : 8.14.0
          colorcet : 3.0.1
           cmocean : v3.0.3
        ipyvtklink : 0.2.3
        ipywidgets : 7.7.5
              tqdm : 4.65.0
            meshio : 5.3.4
         pythreejs : 2.4.2
    pytest_pyvista : 0.1.8
             trame : 2.5.0
      trame_client : 2.9.4
      trame_server : 2.11.4
         trame_vtk : 2.5.0
      nest_asyncio : 1.5.6
--------------------------------------------------------------------------------
adeak commented 1 year ago

@beroda as we discussed in slack, I also see the issue. I checked out e65873eff2c67e83888e9ba22af8e6eceee673e8 (corresponding to #3889 that fixed this), and indeed the issue wasn't there. So this is a regression.

adeak commented 1 year ago

git bisect tells me the regression happened in #4102, looks good based on the name of the PR. cc @akaszynski

adeak commented 1 year ago

Actually, I installed VTK 9.2.2 instead of 9.2.6 (or 9.2.4) and I still see the issue, despite the change in the suspected PR. So that PR might be a red herring on my VTK 9.2.6. Will rerun git bisect...

adeak commented 1 year ago

Yup, with VTK 9.2.2 the issue wasn't fixed by #3889 in the first place.

akaszynski commented 1 year ago

Ran git bisect and came to the same conclusion. Regression introduced in #4102.

adeak commented 1 year ago

@akaszynski I think that PR only reintroduced the regression for VTK 9.2.4 and above, see my previous two comments.

SMoraisAnsys commented 4 months ago

I'm currently facing this error on windows with pyvista==0.44.1 and and vtk==9.3.1. I took the docstring for granted and had a headache to understand why things were not working as expected! If this is the expected behavior for windows, let me know and I can update the docstring accordingly.

Edit: Seems like there is also an issue as I also had the same issue with WSL image