Nothing happens when manimgl is run!

[heethesh]

Describe the error

No output on the console or no GUI shows up when manimgl is run from the command line (installed by pip install manimgl). I observe the same issue when running from this repository and installing using pip install -e . as well.

Code and Error

manimgl

Environment

OS System: Ubuntu 20.04.1 (5.8.0.38-generic)
OpenGL version: 4.6.0 NVIDIA 450.102.04
manim version: 1.0.0 (pip install manimgl)
python version: Tried both 3.6 and 3.8

[Varniex]

You need to pass some flag(s) for it to work.
Try manimgl example_scenes.py OpeningManimExample or manimgl example_scenes.py OpeningManimExample -w to write the scene to a file

[gnap114039]

same issue here !

OS System: Ubuntu 20.04.1 (5.8.0.43-generic)
PyOpenGL: 3.1.5
manim version: 1.0.0 (pip install manimgl)
python version: 3.8.5

I have tried the following command line:

manimgl example_scenes.py OpeningManimExample
manimgl example_scenes.py OpeningManimExample -w
manim-render example_scenes.py OpeningManimExample

the terminal will wait command return after hit enter, and htop showing CPU single thread almost 100 % usage.

[heethesh]

You need to pass some flag(s) for it to work.
Try manimgl example_scenes.py OpeningManimExample or manimgl example_scenes.py OpeningManimExample -w to write the scene to a file

Arguments and flags make no difference, the script should show some error or the help menu when nothing is specified. Works as expected with the ManimCommunity fork. See @gnap114039's comment above. I can see similar CPU usage too.

I'm surprised there are no more reports of this issue, I had the same issue on the master branch even before 1.0.0 was merged/released.

[flyotlin]

Got the same problem, manimgl installed using pip install manimgl.
I've installed ffmpeg and latex before installing manimgl.
No matter arguments or flags are specified or not, nothing pops out in the terminal.

OS: Ubuntu 20.0.1
python: 3.8.5
manim: 1.0.0

[naveen521kk]

Quick answer, try running the command below in same virtual environment.

python -m pip install manimpango --force --no-binary :all:

see if that fixes. Could be related to #1387

[behnamasadi]

Have the same problem, was wondering if it has somethinig to do with OpenGL verson and GPU driver.

[ghost]

Same issue
OS: Ubuntu 20.0.1
python: 3.8.5
manim: 1.0.0

[ghost]

I'm not sure about this
I faced the same problem
but when i install manim using pycharm in venv mode
this issue fixed
something else check the gui driver

[behnamasadi]

I switched to "Manim Community" repo and it worked like a charm, guess thats the one thatI

[ghost]

Downgrade moderngl to 5.5.*
pip install moderngl==5.5.*

[naveen521kk]

@MLee-Sci Can you make an issue in moderngl repo if this fixes?

[einarf]

Do not downgrade moderngl. This can cause more issues.

Downgrading to 5.5.x is not really advicable because it will use statically linked libGL. If it doesn't work with latest moderngl or glcontext it would be nice to figure out why. 😄

[einarf]

EDIT : This is a manimpango issue

Please follow instructions here : #1391 (comment)
(the manimpango README is also helpful)

[heethesh]

python -m pip install -I moderngl==5.5.4 definitely solves it for now

This solves the problem, closing this issue.

[einarf]

The issue itself it not really fixed. It's a band-aid to some other unknown issue. The requirements.txt should ideally be updated to use 5.5.4 until the root cause is identified. It might not even be related to moderngl 5.6.x. Not really sure yet.

[einarf]

We get stuck here when pyglet calls glXChooseFBConfig . Most likely this is related to some memory issue somewhere. The question is where. It could be anything.

#0  0x00007ffff7fa0729 in __pthread_mutex_unlock_usercnt (mutex=mutex@entry=0x5555565a36e8, decr=decr@entry=0) at pthread_mutex_unlock.c:58
#1  0x00007ffff7fa31dd in __pthread_cond_wait_common (abstime=0x0, clockid=0, mutex=0x5555565a36e8, cond=0x5555565a4838) at pthread_cond_wait.c:421
#2  __pthread_cond_wait (cond=0x5555565a4838, mutex=0x5555565a36e8) at pthread_cond_wait.c:638
#3  0x00007fffe931c023 in ?? () from /lib/x86_64-linux-gnu/libxcb.so.1
#4  0x00007fffe931c0d5 in ?? () from /lib/x86_64-linux-gnu/libxcb.so.1
#5  0x00007fffe931c622 in ?? () from /lib/x86_64-linux-gnu/libxcb.so.1
#6  0x00007fffe931c8bb in xcb_send_request_with_fds64 () from /lib/x86_64-linux-gnu/libxcb.so.1
#7  0x00007fffe931cc4d in xcb_send_request () from /lib/x86_64-linux-gnu/libxcb.so.1
#8  0x00007fffe932bfe8 in xcb_query_extension () from /lib/x86_64-linux-gnu/libxcb.so.1
#9  0x00007fffe931eb72 in ?? () from /lib/x86_64-linux-gnu/libxcb.so.1
#10 0x00007fffe931ec2c in xcb_get_extension_data () from /lib/x86_64-linux-gnu/libxcb.so.1
#11 0x00007fffe931c764 in xcb_send_request_with_fds64 () from /lib/x86_64-linux-gnu/libxcb.so.1
#12 0x00007fffe931cc4d in xcb_send_request () from /lib/x86_64-linux-gnu/libxcb.so.1
#13 0x00007fffd2c65b9b in xcb_glx_query_version () from /lib/x86_64-linux-gnu/libxcb-glx.so.0
#14 0x00007fffd2cfda1a in ?? () from /lib/x86_64-linux-gnu/libGLX_mesa.so.0
#15 0x00007fffd2cf9896 in ?? () from /lib/x86_64-linux-gnu/libGLX_mesa.so.0
#16 0x00007fffd2cfb058 in ?? () from /lib/x86_64-linux-gnu/libGLX_mesa.so.0
#17 0x00007fffe4816d6d in glXChooseFBConfig () from /lib/x86_64-linux-gnu/libGLX.so.0
#18 0x00007fffd7e32ff5 in ?? () from /lib/x86_64-linux-gnu/libffi.so.7
#19 0x00007fffd7e3240a in ?? () from /lib/x86_64-linux-gnu/libffi.so.7
#20 0x00007fffd7e4b30e in _call_function_pointer (argcount=4, resmem=0x7fffffff6f70, restype=<optimized out>, atypes=<optimized out>, avalues=0x7fffffff6f40, 

[naveen521kk]

Again, is manimpango the problem? Could you try commenting it out?

[einarf]

@naveen521kk manimgl runs fine if I comment out all manimpango imports. (with moderngl 5.6.4)

.. so we might have an xy problem here.

[naveen521kk]

@einarf Now try, compiling it from source? python -m pip install manimpango --force --no-binary :all: see if the problem occurs with the imports on?

[einarf]

@einarf Now try, compiling it from source? python -m pip install manimpango --force --no-binary :all: see if the problem occurs with the imports on?

Yep. This solves the problem. I'm guessing this is relateed to other segfault issues, but here there was memory currption causing a random issue. The random issue was pyglet hanging on glXChooseFBConfig .. something that is a clear indication there is some curruption going on.

[einarf]

The real solution is to compile manimpango from source. Do not downgrade moderngl 😄

moderngl should be squeeky clean and doesn't really do much on initial import. The action mainly happens on context creation. On import it mainly registers types. I'm guessing a downgrade only changes the memory profile slighty so this random corruption doesn't cause any issues.

[naveen521kk]

I have asked them to do so long back see #1391 (comment)
but nobody cares to read it I think...

I had to have the wheels for Linux simply for readthedocs builds on documentation, which will break if there are no wheels, I have asked users to not install wheels on Linux, but nobody sees to read or it is so hidden...

Meanwhile, let me try if static linking helps here. Again, I would need someone to test or I would simply consider it as fixed :)

[einarf]

@naveen521kk Poke me any time for testing

I should have followed your advice about compiling it, but I got cought up in the debugger 😄

[naveen521kk]

I plan to remove those wheels in the next release see ManimCommunity/ManimPango#27

[MaciejDzikowski]

My solution:
sudo apt install gcc

Then, install the newest manimpango:
pip install manimpango==0.2.6

[Acpear]

Thanks EVERYBODY!!!!!!!! I ! use ! your ! methods ! and ! SUCCEED!!!!! OHHHHHH!!!!!!!! PERFECT !!!!!!!!!