Updated bgfx.cmake to latest.

[bkaradzic-microsoft]

No description provided.

[Copilot]

Pull request overview

Updates the BabylonNative build’s FetchContent pin for bgfx.cmake to a newer upstream commit, keeping the dependency set current while maintaining the project’s existing “pin by commit SHA” approach for most third-party sources.

Changes:

• Bumped bgfx.cmake GIT_TAG from 3b67f0c0c5505a22f6e88640519b5b473c938ad9 to 44adc64c885739d8721ab97fb7bd511ff72e26d6.

[bkaradzic-microsoft]

CI status on the latest bgfx.cmake bump: everything green except iOS_Xcode26 and iOS_Installation, which both fail with the same single error:

clang: error: argument unused during compilation: '-Xarch_x86_64 -msse4.2'
       [-Werror,-Wunused-command-line-argument]

This is a gap in the arch-aware SSE minspec that landed in BabylonJS/bgfx.cmake#133. That change uses -Xarch_x86_64 for universal Apple builds so the flag only reaches the x86_64 slice, which is correct when CMAKE_OSX_ARCHITECTURES describes the slices actually compiled -- true for macOS universal (those jobs pass), but not for Xcode in general.

Our iOS configure goes through leetal/ios-cmake with PLATFORM=OS64COMBINED, which reports architecture(s): arm64;x86_64. Xcode picks slices per SDK at build time, and the Release-iphoneos build compiles arm64 only (x86_64 is the simulator slice, built separately). So the -Xarch_x86_64 applies to nothing, and -Werror turns the resulting unused-argument warning into a build failure.

Fix is up as BabylonJS/bgfx.cmake#134: pair the flag with -Wno-unused-command-line-argument, scoped to the universal-build branch only, so every x86_64 slice that is really compiled still gets the SSE4.2 minspec. I'll bump the GIT_TAG here once that merges.

[bkaradzic-microsoft]

Update: the run finished at 31/33 green -- iOS_Xcode26 and iOS_Installation are the only failures, and both are the single -Xarch_x86_64 error described above. Everything else, including all the macOS, UWP arm64, Linux, Android and Win32 jobs, passed on the new bgfx.cmake tag.

I've verified the fix (BabylonJS/bgfx.cmake#134) end-to-end rather than guessing: I pointed a throwaway branch's GIT_TAG at that commit and re-ran just these two jobs on the same runner image and Xcode.

Job	bgfx.cmake 7062f11	bgfx.cmake d8e3ba1 (#134)
iOS_Xcode26 / build	❌ exit 65	✅ 4m4s
iOS_Installation / test-install	❌ exit 65	✅ 3m14s

So #134 is the only thing standing between this PR and a fully green run. Once it merges I'll bump the GIT_TAG here and re-run.

[bghgary]

[Reviewed by Copilot on behalf of @bghgary]

LGTM

[bkaradzic-microsoft]

Green across the board -- 33/33 on 6c55158 (BabylonJS/bgfx.cmake#134). iOS_Xcode26 and iOS_Installation, the only two failures on the previous tag, both pass now, and nothing else regressed.