https://bugzilla.redhat.com/show_bug.cgi?id=2134911
Bug ID: 2134911 Summary: QSocketNotifier: Can only be used with threads started with QThread Product: Fedora Version: 36 Hardware: x86_64 OS: Linux Status: NEW Component: python-matplotlib Assignee: quantum.analyst@gmail.com Reporter: dev@javinator9889.com QA Contact: extras-qa@fedoraproject.org CC: epel-packagers-sig@lists.fedoraproject.org, gwync@protonmail.com, jonathan.underwood@gmail.com, paulo.cesar.pereira.de.andrade@gmail.com, python-packagers-sig@lists.fedoraproject.org, quantum.analyst@gmail.com, thibault@north.li, tomspur@fedoraproject.org Target Milestone: --- Classification: Fedora
Created attachment 1918084 --> https://bugzilla.redhat.com/attachment.cgi?id=1918084&action=edit Python script to replicate exactly the issue (one needs to have cuDNN and CUDA for make it work)
Description of problem:
Trying to replicate a convolutional AI model on my local computer I faced this issue when simply trying to display the generated images by such model. The "guide" I'm following is: https://keras.io/guides/keras_cv/generate_images_with_stable_diffusion/?s=09...
There is no issue when generating the images but when displaying them.
Version-Release number of selected component (if applicable): tried with both matplotlib from pip (3.6.1) and the one bundled with Fedora 36 (3.5.3-2, at the time of writting this). Notice that the shipped Python version is in use (3.10.2)
How reproducible: always, inside a Wayland environment (I didn't tried with Xorg but I've seen in general that this issue is only related to Wayland)
Steps to Reproduce: 1. Generate a bunch of images using the model, or any other application 2. In a Python console: ``` import matplotlib.pyplot as plt
# images = <matrix of generated images> plt.imshow(images) ```
Actual results: QSocketNotifier: Can only be used with threads started with QThread
Expected results: The set of images show up and are displayed