Skip to content

Fix invalid source_group arguments in Canvas CMakeLists - #2093

Merged
bkaradzic-microsoft merged 1 commit into
BabylonJS:masterfrom
bkaradzic-microsoft:fix-canvas-source-group
Nov 18, 2026
Merged

bkaradzic-microsoft merged 1 commit into
BabylonJS:masterfrom
bkaradzic-microsoft:fix-canvas-source-group

Conversation

@bkaradzic-microsoft

Copy link
Copy Markdown
Member

source_group(<name> FILES ...) doesn't accept a directory argument. The ubuntu-26.04 runner image 20261116.202.1 ships a newer CMake, which now rejects the stray ${CMAKE_CURRENT_SOURCE_DIR} in Polyfills/Canvas/CMakeLists.txt:

CMake Error at Polyfills/Canvas/CMakeLists.txt:70 (source_group):
  source_group given unknown argument:
  "/home/runner/work/BabylonNative/BabylonNative/Polyfills/Canvas".

This breaks Linux configure for any new run on that image (first seen on #2092 Ubuntu_Clang_JSC). This PR removes the extra argument. IDE grouping is unchanged, because the grouping comes from the group name plus FILES.

source_group(<name> FILES ...) does not take a directory argument. Newer
CMake on the ubuntu-26.04 runner image (20261116) rejects the stray
${CMAKE_CURRENT_SOURCE_DIR} with "source_group given unknown argument",
failing configure on Linux.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 8bd57bfd-bbb3-4e86-8a17-9a44fb995d0f

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟢 Approval recommended

No active review issues remain for the Canvas CMake configuration change.

No open issues 🎉

What changed in this PR

Fixes Canvas configuration failures with newer CMake versions while preserving IDE source grouping.

Changes:

  • Removes the unsupported directory argument from the third-party and shader source_group calls, retaining their names and file lists.
File Description
Polyfills/​Canvas/​CMakeLists.txt Corrects arguments in two source_group calls.

Review effort: Balanced

@bkaradzic-microsoft

Copy link
Copy Markdown
Member Author

CI note (run 44339706050, attempt 1)

Two macOS jobs were cancelled without ever getting a runner. This is infra, not this PR:

Job Queue wait Result
iOS_Installation / test-install 15:47:36Z → 16:13:19Z, 25m43s, no runner cancelled
iOSTests / test 16:01:06Z → 16:19:50Z, 18m44s, no runner cancelled

All other jobs passed, including Ubuntu_Clang_JSC on the new ubuntu-26.04 image (queued 2s, ran 20m28s). That confirms the fix. Other macOS jobs did get runners: MacOS_Ninja waited 11m01s and ran 7m53s; MacOS waited 4m09s and ran 14m51s. The longest job was Win32_x64_V8_D3D11, which waited 3s and ran 37m55s.

Total wall-clock: 15:47:35Z → 16:25:47Z = 38m12s.

Rerunning the failed jobs.

@bkaradzic-microsoft

Copy link
Copy Markdown
Member Author

CI note (run 44339706050, attempt 2)

iOSTests / test was cancelled again without ever getting a macOS runner. This is infra, not this PR:

Job Queue wait Run time Result
iOS_Installation / test-install 5s 6m17s success
iOSTests / test 16:49:28Z → 17:04:30Z, 15m02s, no runner n/a cancelled

Attempt 2 wall-clock: 16:49:22Z → 17:04:39Z = 15m17s. Total since the first attempt: 15:47:35Z → 17:04:39Z = 1h17m04s.

Every other job passed in attempt 1, including Ubuntu_Clang_JSC on the new ubuntu-26.04 image. Rerunning the failed jobs.

@bghgary bghgary left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[Reviewed by Copilot on behalf of @bghgary]

LGTM.

@bkaradzic-microsoft
bkaradzic-microsoft merged commit 1662fc8 into BabylonJS:master Nov 18, 2026
87 of 92 checks passed
bkaradzic-microsoft added a commit that referenced this pull request Nov 19, 2026
## Summary
- bump Babylon.js packages from 9.35.0 to 9.37.1
- enable the deferred post-`setStateCullFaceType` pixel expectations
from #2069 now that BabylonJS/Babylon.js#19175 is released
- re-enable `Gaussian Splatting Loading RH` and `Gaussian Splatting LOD`
after BabylonJS/Babylon.js#19182
- narrow 18 mesh-blending, object-ID, and FrameGraph geometry exclusions
to D3D11/D3D12 after BabylonJS/Babylon.js#19150 and #19151 (six of them
also stay excluded on OpenGL: the FrameGraph geometry renderer scenes
fail to compile the GPU particle update effect, and the coastal cliff
and Gaussian mesh scenes time out in createScene on Linux CI); polyfill
`Object.fromEntries` for ChakraCore in `validation_native.js` (next to
the existing ES2019 polyfills) and re-enable the four FBX animation
scenes
- enable `Alpha-to-coverage transparent overlap` now that #2082 is
merged and Babylon.js 9.37.1 includes BabylonJS/Babylon.js#19148

Rebased onto master after #2059, #2082 and #2093 merged.

## Validation
- Win32 RelWithDebInfo `UnitTests`: 200/200 passed (JavaScript: 150
passed, 29 pending)
- released-fix candidates: Gaussian Splatting Loading RH and LOD passed;
alpha-to-coverage passes on Win32 Chakra D3D11 after #2082 (0.229% pixel
difference); compound splat and FBX tests remain excluded for
independent runtime blockers
- full D3D11 Playground sweep: 757 passed, 78 failed, 57 skipped versus
master baseline 752 passed, 81 failed, 59 skipped; no new failures
- Linux OpenGL CI (Ubuntu_Clang_JSC, run 44337090420 attempt 2): 725
passed, 6 failed before those six were excluded on OpenGL
- Win32 Chakra D3D11: all 18 `FBX loader` tests pass; without the
polyfill `FBX loader - skeletal animation` fails with `Object doesn't
support property or method 'fromEntries'`

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 8bd57bfd-bbb3-4e86-8a17-9a44fb995d0f
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants