Skip to main content

project-setup-python

Configure distributed OVXR Python bindings and their native dependencies. Use for Python SDK setup, loader failures, or package capability checks.

Source facts

Repository
NVIDIA-Omniverse/ovxr
Last source activity
October 20, 2026 at 13:19
Detected SKILL.md language
English
Stars
10
Forks
0

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
project-setup-python
description
Configure distributed OVXR Python bindings and their native dependencies. Use for Python SDK setup, loader failures, or package capability checks.
license
LicenseRef-NvidiaProprietary
metadata
{"author":"mbickleyNV <129252267+mbickleyNV@users.noreply.github.com>","version":"0.1.0","tags":["ovxr","openxr","sdk-integration"],"domain":"graphics"}
# Project Setup Python ## Requirements Requires a matched distributed OVXR, OVRTX and OVStage SDK and a supported OpenXR runtime. No Kit or implementation-source checkout is required. ## Instructions Use a matched OVXR, OVRTX and OVStage package set. Record platform, archive/build identity and loaded module paths; an unchanged OVXR version string does not establish ABI compatibility. The authoritative public README specifies a bindings-only Python wheel: supply a matching native OVXR runtime and its OVRTX dependencies separately. OVRTX and OVStage must remain matched. Artifact-specific packaging observations do not replace that documented deployment contract. Select an OpenXR runtime with process-local `XR_RUNTIME_JSON`. For OVXR loading, set `ovxr.OVXR_LIBRARY_PATH_HINT` to a shared-library file or containing directory for an intentional override; an unusable hint fails without fallback. Otherwise discovery checks the package-local bin, owning deploy bin and plugin directories, then the linked build plugin directory. Default discovery does not search PATH or LD_LIBRARY_PATH. Development-only system and current-directory searches require their separate documented opt-ins. On Windows, a split deployment must put its trusted bin directory on the process native PATH for by-name payload dependencies; this is separate from Python shim discovery. On Linux OVXR preloads matching OVRTX with RTLD_GLOBAL. Follow OVRTX's own hint/candidate policy or reuse its loaded library so both bindings share one OVRTX instance. Do not change machine-wide loader or OpenXR settings. OVXR preloads its borrowed runtime dependencies. Use the package's documented loader configuration; preserve its ABI and exported interfaces. Renderer-global settings, when needed, must be applied before renderer construction. Validate in separate gates: Python import, matching native exports, stage population, runtime/session readiness, then a submitted image. Import success does not prove plugins, shaders, graphics support or a connected client. For lifecycle read [application-flow](../application-flow/SKILL.md); for readiness and evidence read [validation](../validation/SKILL.md). ## Examples For a supplied Python SDK, install its matching platform wheel in an isolated environment. Record package versions and imported module locations, then separate import success from native-load and first-frame results.
View on GitHub