slang.git/.github, branch master

Re-enable slangpy test_blit.py::test_generate_mips for CUDA. (#8740)

2025-10-17T00:37:51+00:00

Issue is fixed with https://github.com/shader-slang/slang/pull/8710

Update slang-rhi (#8709)

2025-10-16T17:18:31+00:00

- Update to latest slang-rhi - Enable additional slang-rhi tests for OptiX 8.0 and 8.1

[CI] Skip slangpy test_blit.py::test_generate_mips for CUDA (#8725)

2025-10-16T00:03:56+00:00

This skips a new test from slangpy that is hitting an internal assert in slang CI, uncaught in testing due to slangpy's CI testing using release builds. See https://github.com/shader-slang/slangpy/issues/575 for details

Defer `IRCastStorageToLogicalDeref` in lowerBufferElementType pass. (#8668)

2025-10-10T01:30:24+00:00

Fix a regression on metal test. In `lowerBufferElementTypeToStorageType` pass, not only we want to defer an argument that is `CastStorageToLogical` to the callee, but also apply the same defer logic to `CastStorageToLogicalDeref` as well. Because `CastStorageToLogicalDeref` will appear as argumnet if `lowerBufferElementTypeToStorageType` is run before we apply the `in->borrow` transformation pass, which is the case for metal parameter block legalization.

Use GitHub runners for Windows releases, disable CUDA for aarch64 (#8613)

2025-10-07T08:56:23+00:00

For #8596 Fixes #8597 This switches our release workflow back to using GitHub's `windows-latest` runners, which we were using previously. It also adds the variable `extra-cmake-flags` to the `windows-aarch64` entry in the workflow's matrix with the value `"-DSLANG_ENABLE_CUDA=0"`. If we are cross-compiling aarch64 on x86_64, and the x86_64 CUDA Toolkit is installed, it will be auto-detected by cmake and the build will fail (no aarch64 version of CUDA Toolkit exists). The `windows-latest` runners do not have CUDA Toolkit, so they do not encounter this issue, but if we do end up building on runners that do (such as the temporary move to self-hosted runners), adding that flag eliminates that potential problem. This release workflow does build properly on `windows-latest` with `extra-cmake-flags`: https://github.com/aidanfnv/slang/actions/runs/18293521738

Re-add macOS cross-compile setup to release workflow (#8583)

2025-10-01T22:58:40+00:00

For #8564 Similar to #8580, this re-adds the cross-compile target setup step for macOS releases that was erroneously removed in https://github.com/shader-slang/slang/pull/8470, which made x86_64 releases build aarch64 binaries. It also simplifies the workflow logic a bit by adding a separate `arch` variable to the release matrix, which refers to the target architecture in the manner that the setup requires, so that we do not have to replace the string `"aarch64"` with `"arm64"` in setting `CMAKE_OSX_ARCHITECTURES` for native aarch64 macOS builds and do not have to conditionally set the MSVC `arch` to `amd64_arm64` for Windows cross-compilation. --------- Co-authored-by: slangbot Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>

Re-add Windows cross-compile setup from release workflow (#8580)

2025-10-01T20:07:55+00:00

For #8578 This re-adds the cross-compile target setup step for Windows releases that was erroneously removed in #8470, which made aarch64 releases build x64 binaries. The flow should be: - setup MSVC for host arch - build generators with cmake - setup MSVC for cross-compile target arch - build slang Based on the description of #8470, it seems that the cross-compile MSVC step was mistaken as a duplicate for the host MSVC step and removed for being seemingly redundant.

Add timeout to CI jobs; add mimalloc to .gitignore (#8538)

2025-09-25T00:54:26+00:00

Adding slang-test option to ignore abort popup message (#8492)

2025-09-24T06:46:31+00:00

With the recent Windows runtime libraries, a new popup window started appearing when `abort()` is called. This was observed when VVL prints a message as a part of WGPU test. Although it can be helpful when we want to debug it, it breaks the behavior of CI scripts when the tests are expected to continue even when they fail. When the test fail, CI script stops in the middle and wait for a user to click on a button on the dialog window, which cannot happen. As a result, when there is a VVL error message, CI run stops in the middle and the testing stops prematurely. This commit adds a new command-line argument, `-ignore-abort-msg`, that ignores the abort message and it wouldn't show the dialog popup window. From the implementation perspective, there are three places that are related. - slang-test itself should turn off the flag. - render-test should turn off the flag after getting the argument from slang-test - test-server should turn off the flag after getting the argument from slang-test When test-server runs render-test, the arguments are already handled by slang-test, so test-server needs to just pass through the arguments.

Reduce the maxprocesses to 3 for slangpy test (#8502)

2025-09-22T22:28:22+00:00