napari_track_edit._wgpu_setup
=============================

.. py:module:: napari_track_edit._wgpu_setup

.. autoapi-nested-parse::

   wgpu instance configuration, applied before fastplotlib/pygfx create an instance.



Attributes
----------

.. autoapisummary::

   napari_track_edit._wgpu_setup.logger


Functions
---------

.. autoapisummary::

   napari_track_edit._wgpu_setup.configure_wgpu_backend


Module Contents
---------------

.. py:data:: logger

.. py:function:: configure_wgpu_backend() -> None

   Restrict the wgpu instance to Vulkan on Linux.

   wgpu probes every backend when requesting an adapter. On Linux the GLES/EGL probe
   calls eglMakeCurrent, which returns BadAccess once napari's Qt/vispy OpenGL context
   is current; wgpu-native then panics in non-unwinding FFI code and aborts the whole
   process. Mesa ships the lavapipe ICD as a software fallback, so Vulkan-only is safe.

   Set NAPARI_TRACK_EDIT_WGPU_BACKENDS (comma-separated, e.g. "All") to override.