[CedricGuillemet]

Describe the change

Prepare the stable 2.0.8 release with the NativeXr frame-pairing fix from BabylonJS/BabylonNative#1957.

Package/gulpfile.js pins the immutable upstream Babylon Native commit 37d30f54649543290ad25c6c3626c247c1372773. The fix queues EndFrame during FrameStart so it cannot miss its own AfterRender tick and consume the next frame, which caused the m_sessionState->Frame != nullptr assertion. The dependency comes from BabylonJS/BabylonNative, not the earlier fork-only fix or the intentionally failing repro branch.

Update both READMEs from preview framing to the planned stable 2.0.8 release, documenting the exact native dependency and unchanged compatibility baselines. Existing stable 2.0.7 installation guidance remains until 2.0.8 is published. Package version metadata follows the existing publish workflow, which sets the requested version at publication time.

Release-preparation commit 201bb65e240b2239a32d365c406ed4ce879962e1 is pushed to both the fork PR branch and the upstream branch. No package publication, release/tag creation, or publishing-workflow dispatch was performed as part of this update.

Screenshots

N/A - dependency pin and documentation changes only.

Documentation

Updated README.md and Modules/@babylonjs/react-native/README.md for the planned stable 2.0.8 release, NativeXr fix, and unchanged compatibility requirements.

Testing

• PR Build run 39740666984 passed all eight jobs: TypeScript, package generation, Android/iOS/Windows source builds, and Android/iOS/Windows packaged-consumer builds. This run tested dependency-pin commit f661dbb28f9c35cf7730355e8d0f98f2e904cd4e; subsequent commits change READMEs only.
• Verified the staging artifact's NativeXrImpl.cpp matches Git blob e8dcade6be34c21305a3bf1b8f1ca2ad97fa40d5 at the pinned Babylon Native commit.
• Clean-built ARM64 ARViewTest from that staging artifact with app/native Debug, BX_CONFIG_DEBUG=1, and no NDEBUG, then installed it on Pixel 9a. A six-minute loaded-AR soak at approximately 30 FPS and three additional Place Media/Back cycles passed in the same process, with no fatal or assertion logs. Model loading and placement were recorded. Debug re-entry remained 30-32 seconds; Back was observed within 4.3-4.8 seconds including UI-sampling overhead.
• Separately installed the published npm @babylonjs/react-native@2.0.8-preview, clean-built app/native Debug, and successfully launched it on Pixel 9a. Its XR source matches the same pinned fix. The six-minute soak was on the CI staging artifact, not repeated on the published preview.
• node --check Package/gulpfile.js and git diff --check passed during the dependency/documentation updates. No full native rebuild was performed for the README-only release-preparation change. Device coverage is bounded lifecycle testing, not the complete manual Android/iOS Playground checklist.

[Copilot]

🟢 Approval recommended

The narrowly scoped pin update has reported passing cross-platform CI, the remaining changes are documentation-only, and no blocking issues were identified.

No open issues 🎉

What changed in this PR

Stages the merged Babylon Native NativeXr EndFrame fix for the 2.0.8 preview target while keeping 2.0.7 identified as the latest stable release.

Changes:

• Pins Babylon Native packaging to immutable upstream commit 37d30f54649543290ad25c6c3626c247c1372773.
• Documents the preview target, lifecycle fix, and unchanged compatibility baselines.
• Leaves package version metadata and build infrastructure unchanged.

File	Description
README.md	Introduces the 2.0.8 preview target and summarizes the fix and compatibility baselines.
Package/​gulpfile.js	Updates the Babylon Native source archive pin to the merged fix.
Modules/​@babylonjs/​react-native/​README.md	Documents preview publishing expectations, the source pin, and supported versions.

_{Review effort: Balanced}

[ryantrem]

I'm a bit confused why we want to check in doc changes referencing upcoming changes.

[bghgary]

[Reviewed by Copilot on behalf of @bghgary]

LGTM