[hybrid] Hybrid API: native library tasks with cuDNN provider by mikepapadim · Pull Request #888 · beehive-lab/TornadoVM

mikepapadim · 2026-07-02T20:11:14Z

Description

Adds the cuDNN provider to the Hybrid API — CNN primitives plus fused scaled-dot-product attention (flash attention, FP16) via the cuDNN graph API.

Stacked on #887 (this branch includes its commits: the library-task SPI, cuBLAS/cuBLASLt, and cuFFT providers). Suggest reviewing #887 first and re-targeting this PR's base to develop for consistency.

What's included

Legacy-API primitives (FP32, NCHW) — tornado-cudnn + tornado-drivers/cudnn-jni (requires cuDNN 9):

Factory	cuDNN call	Semantics
`cudnnSoftmax(in, out, rows, cols)`	`cudnnSoftmaxForward` (ACCURATE/INSTANCE)	numerically stable per-row softmax
`cudnnRelu` / `cudnnSigmoid` / `cudnnTanh`	`cudnnActivationForward`	element-wise activations
`cudnnMaxPool2d`	`cudnnPoolingForward` (MAX)	square window/stride
`cudnnConv2d`	`cudnnConvolutionForward`	2D cross-correlation, NCHW/KCRS; per-shape plan (descriptors + algorithm + workspace size) with a grow-only device workspace

These legacy functions are deprecated-since-cuDNN-9.0 (fully functional, validated on 9.23); tornado-cudnn/ROADMAP.md tracks their migration to graph-API patterns behind the same factories.

Fused SDPA / flash attention (the flagship): CuDnn.sdpaForward(q, k, v, o, b, h, sQ, sKv, d, scale, causal) — FP16 HalfFloatArrays in packed BHSD layout, FP32 accumulation, inference, optional causal mask (d multiple of 8, ≤ 256, Ampere+). Implemented with the header-only cudnn-frontend v1.25.0 (NVIDIA-recommended over raw backend descriptors), fetched at build time; graph → heuristic → execution plan bundled per shape and created in the provider's prepare() hook, so SDPA is safe with CUDA graph capture.

Core fix: TornadoTaskGraph.isArgumentIgnorable now covers Boolean, HalfFloatArray, and ByteArray task arguments — a Boolean task argument previously crashed lockInPendingFieldsObjects with IndexOutOfBoundsException (surfaced by the SDPA causal flag; latent for the array types).

Tests

uk.ac.manchester.tornado.unittests.cudnn.TestCuDnn — 9/9 via tornado-test (auto-skip [UNSUPPORTED] without CUDA backend / libtornado-cudnn): softmax/relu/tanh/maxpool/conv2d vs Java references; a CNN mixed pipeline (conv → JIT bias task → ReLU → max pool on shared device buffers); conv under withCUDAGraph; SDPA plain and causal vs a Java attention reference. Cross-suite regression after the core fix: cuBLAS 11/11, cuBLASLt 5/5, cuFFT 8/8, TestArrays 23/23.

Benchmarks (RTX 4090, cuDNN 9.23, cross-validated)

Convolution (BenchmarkConv2d, FP32):

Layer (NCHW, 3×3, pad 1)	TornadoVM JIT direct conv	cuDNN	Speedup
8×64×56×56, k=64	3.8 TFLOP/s	11.6 TFLOP/s	3.1×
16×128×28×28, k=128	4.0 TFLOP/s	21.1 TFLOP/s	5.3×

Fused causal flash attention (BenchmarkSdpa, FP16 vs the FP32 JIT attention kernel on identical FP16-rounded inputs):

Shape (b, h, s, d)	JIT attention kernel	cuDNN fused SDPA	Speedup
4, 16, 1024, 128	90.7 ms	0.268 ms (64.2 TFLOP/s)	339×
1, 32, 2048, 128	203.3 ms	0.407 ms (84.3 TFLOP/s)	499×

Docs

tornado-cudnn/README.md (supported ops, deprecation note, benchmark tables) and tornado-cudnn/ROADMAP.md (coverage tracker derived from the cuDNN docs: legacy completion, FP16 Tensor Core conv + autotuning, RMSNorm/LayerNorm, GQA/paged-KV attention, graph-API migration; FP8 items marked hardware-gated on Hopper).

Backend/s tested

Verified on the CUDA backend (make BACKEND=cuda), RTX 4090; suites skip cleanly on other backends.

Copilot

Pull request overview

Adds a “Hybrid API” execution path so a TaskGraph can mix JIT-compiled kernels with calls into native vendor libraries (cuBLAS/cuFFT/cuDNN) on the same device buffers/stream, plus new CUDA library modules, JNI drivers, tests, docs, and packaging updates to ship/run them.

Changes:

Introduce library-task descriptors + runtime SPI (TornadoLibraryProvider) and runtime plumbing to create per-(device, plan) native contexts and dispatch library calls from the interpreter.
Expose CUDA native stream/context handles for ordering native library work with TornadoVM kernels/transfers (incl. CUDA Graph capture constraints).
Add new modules (tornado-cublas, tornado-cufft, tornado-cudnn), JNI driver submodules, unit tests/benchmarks, documentation, and distribution wiring.

Reviewed changes

Copilot reviewed 89 out of 91 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
tornado-unittests/src/main/java/uk/ac/manchester/tornado/unittests/cufft/TestCuFft.java	Adds cuFFT unit tests using library tasks with a Java DFT reference.
tornado-unittests/src/main/java/module-info.java	Adds module requires/exports for new CUDA library test packages.
tornado-unittests/pom.xml	Adds dependencies on `tornado-cublas`, `tornado-cufft`, `tornado-cudnn`.
tornado-runtime/src/main/java/uk/ac/manchester/tornado/runtime/tasks/TornadoTaskGraph.java	Extends task-graph runtime to accept/track library tasks and cleanup contexts.
tornado-runtime/src/main/java/uk/ac/manchester/tornado/runtime/tasks/LibraryTask.java	Introduces a schedulable task type representing a native library call.
tornado-runtime/src/main/java/uk/ac/manchester/tornado/runtime/library/spi/TornadoNativeStreamSupport.java	Defines backend capability to expose native stream/context handles.
tornado-runtime/src/main/java/uk/ac/manchester/tornado/runtime/library/spi/TornadoLibraryProvider.java	Adds provider SPI for native library bindings.
tornado-runtime/src/main/java/uk/ac/manchester/tornado/runtime/library/spi/LibraryInvocation.java	Defines the runtime-resolved argument/pointer bundle passed to providers.
tornado-runtime/src/main/java/uk/ac/manchester/tornado/runtime/library/spi/LibraryContext.java	Defines per-(library, device, plan) opaque provider context.
tornado-runtime/src/main/java/uk/ac/manchester/tornado/runtime/library/LibraryRegistry.java	Adds provider discovery and context caching/teardown registry.
tornado-runtime/src/main/java/uk/ac/manchester/tornado/runtime/interpreter/TornadoVMInterpreter.java	Adds interpreter support to prepare and dispatch library tasks at launch time.
tornado-runtime/src/main/java/module-info.java	Exports/uses library SPI packages from `tornado.runtime`.
tornado-drivers/pom.xml	Adds new JNI driver submodules under the CUDA profile.
tornado-drivers/cufft-jni/src/main/cpp/source/tornado-cufft.cpp	Implements JNI wrappers around cuFFT plan/exec/stream binding.
tornado-drivers/cufft-jni/src/main/cpp/CMakeLists.txt	Builds `tornado-cufft` JNI shared library and locates cuFFT.
tornado-drivers/cufft-jni/pom.xml	Maven build for cuFFT JNI driver (cmake + packaging).
tornado-drivers/cudnn-jni/src/main/cpp/source/tornado-cudnn-sdpa.cpp	Adds cuDNN graph-API SDPA (flash attention) JNI implementation.
tornado-drivers/cudnn-jni/src/main/cpp/CMakeLists.txt	Builds `tornado-cudnn` JNI shared library and fetches cudnn-frontend headers.
tornado-drivers/cudnn-jni/pom.xml	Maven build for cuDNN JNI driver (cmake + packaging).
tornado-drivers/cuda/src/main/java/uk/ac/manchester/tornado/drivers/cuda/runtime/CUDATornadoDevice.java	Implements `TornadoNativeStreamSupport` for CUDA devices.
tornado-drivers/cuda/src/main/java/uk/ac/manchester/tornado/drivers/cuda/CUDADeviceContextInterface.java	Adds default native stream/context interop hooks to device contexts.
tornado-drivers/cuda/src/main/java/uk/ac/manchester/tornado/drivers/cuda/CUDADeviceContext.java	Exposes CUstream/CUcontext handles from CUDA command queues.
tornado-drivers/cuda/src/main/java/uk/ac/manchester/tornado/drivers/cuda/CUDACommandQueue.java	Adds Java natives to fetch raw stream/context pointers.
tornado-drivers/cuda-jni/src/main/cpp/source/CUDAGraph.cpp	Implements JNI natives to return queue CUstream/CUcontext pointers.
tornado-drivers/cuda-jni/src/main/cpp/CMakeLists.txt	Adjusts CUDA include/lib hint ordering for consistent toolkit selection.
tornado-drivers/cublas-jni/src/main/cpp/CMakeLists.txt	Adds build for `tornado-cublas` JNI shared library.
tornado-drivers/cublas-jni/pom.xml	Maven build for cuBLAS JNI driver (cmake + packaging).
tornado-cufft/src/main/resources/META-INF/services/uk.ac.manchester.tornado.runtime.library.spi.TornadoLibraryProvider	Registers cuFFT provider via ServiceLoader.
tornado-cufft/src/main/java/uk/ac/manchester/tornado/cufft/tests/BenchmarkFft.java	Adds FFT benchmark comparing Java/JIT DFT vs cuFFT library task.
tornado-cufft/src/main/java/uk/ac/manchester/tornado/cufft/provider/CuFftNativeLib.java	Adds JNI loader + cuFFT native method declarations and error decoding.
tornado-cufft/src/main/java/uk/ac/manchester/tornado/cufft/provider/CuFftLibraryProvider.java	Implements cuFFT provider and plan caching per shape/context.
tornado-cufft/src/main/java/uk/ac/manchester/tornado/cufft/CuFft.java	Adds public factory methods for cuFFT library task descriptors.
tornado-cufft/src/main/java/module-info.java	Declares `tornado.cufft` module and provides provider implementation.
tornado-cufft/README.md	Documents cuFFT library task usage and build/run instructions.
tornado-cufft/pom.xml	Maven module for cuFFT Java bindings.
tornado-cudnn/src/main/resources/META-INF/services/uk.ac.manchester.tornado.runtime.library.spi.TornadoLibraryProvider	Registers cuDNN provider via ServiceLoader.
tornado-cudnn/src/main/java/uk/ac/manchester/tornado/cudnn/tests/BenchmarkSdpa.java	Adds SDPA benchmark (JIT vs fused cuDNN graph API).
tornado-cudnn/src/main/java/uk/ac/manchester/tornado/cudnn/tests/BenchmarkConv2d.java	Adds conv2d benchmark (JIT vs cuDNN).
tornado-cudnn/src/main/java/uk/ac/manchester/tornado/cudnn/provider/CuDnnNativeLib.java	Adds JNI loader + native method declarations for cuDNN ops and plans.
tornado-cudnn/src/main/java/uk/ac/manchester/tornado/cudnn/CuDnn.java	Adds public factory methods for cuDNN library task descriptors.
tornado-cudnn/src/main/java/module-info.java	Declares `tornado.cudnn` module and provides provider implementation.
tornado-cudnn/README.md	Documents cuDNN library task usage, requirements, and benchmarks.
tornado-cudnn/pom.xml	Maven module for cuDNN Java bindings.
tornado-cublas/src/main/resources/META-INF/services/uk.ac.manchester.tornado.runtime.library.spi.TornadoLibraryProvider	Registers cuBLAS and cuBLASLt providers via ServiceLoader.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/TestCuBlasWorkspace.java	Adds cuBLAS workspace allocation/reuse test.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/TestCuBlasSgemvWithTornadoVMTasksPOST.java	Adds mixed JIT + cuBLAS + JIT pipeline test.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/TestCuBlasSgemvWithTasksCudaGraph.java	Adds CUDA Graph capture/replay integration test for mixed graphs.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/TestCuBlasSgemvBeta.java	Adds SGEMV beta!=0 test validating READ_WRITE semantics.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/TestCuBlasSgemv.java	Adds basic SGEMV correctness test.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/TestCuBlasSgemmTF32.java	Adds TF32-vs-FP32 SGEMM validation + timing.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/TestCuBlasSgemm.java	Adds SGEMM correctness test using row-major trick.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/TestCuBlasGemmExFP16.java	Adds mixed-precision GEMMEx FP16 tests.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/tests/BenchmarkLtFusedMlp.java	Adds cuBLASLt fused MLP benchmark (GELU_BIAS epilogue).
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/provider/CuBlasNativeLib.java	Adds JNI loader + native method declarations for cuBLAS calls.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/provider/CuBlasLtNativeLib.java	Adds JNI declarations for cuBLASLt handle/plan/execute.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/provider/CuBlasLtLibraryProvider.java	Implements cuBLASLt provider with plan cache and workspace management.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/enums/CudaDataType.java	Adds CUDA data type enum mapping for Ex APIs.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/enums/CuBlasOperation.java	Adds cuBLAS transpose-op enum mapping.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/enums/CuBlasMathMode.java	Adds cuBLAS math-mode enum mapping (incl. TF32).
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/enums/CuBlasLtEpilogue.java	Adds cuBLASLt epilogue enum mapping (bias/GELU).
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/enums/CublasComputeType.java	Adds cuBLAS compute-type enum mapping.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/CuBlasOptions.java	Adds per-call tuning options container for cuBLAS tasks.
tornado-cublas/src/main/java/uk/ac/manchester/tornado/cublas/CuBlasLt.java	Adds public factory methods for cuBLASLt library task descriptors.
tornado-cublas/src/main/java/module-info.java	Declares `tornado.cublas` module and provides provider implementations.
tornado-cublas/pom.xml	Maven module for cuBLAS Java bindings.
tornado-assembly/src/bin/tornado.py	Adds CUDA library modules to `--add-modules` when CUDA backend is enabled.
tornado-assembly/src/bin/tornado-test	Adds cuBLAS/cuFFT/cuDNN tests to the “test the world” list.
tornado-assembly/pom.xml	Adds assembly dependencies for CUDA library modules + JNI libs.
tornado-assembly/assembly.xml	Includes new CUDA library modules and JNI artifacts in the distribution.
tornado-api/src/main/java/uk/ac/manchester/tornado/api/TornadoTaskGraphInterface.java	Adds API hook for adding a library task to a task graph.
tornado-api/src/main/java/uk/ac/manchester/tornado/api/common/TornadoFunctions.java	Adds functional interfaces for `libraryTask(...)` factories.
tornado-api/src/main/java/uk/ac/manchester/tornado/api/common/LibraryTaskDescriptor.java	Introduces the public descriptor type for a library task call.
pom.xml	Adds new top-level Maven modules for cuBLAS/cuFFT/cuDNN.
docs/source/index.rst	Adds the hybrid API doc page to the docs index.
docs/source/hybrid-api.rst	Documents hybrid API concepts, usage, and provider SPI.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

+    public static void destroyContexts(long executionPlanId) {
+        Iterator<Map.Entry<ContextKey, LibraryContext>> iterator = CONTEXTS.entrySet().iterator();
+        while (iterator.hasNext()) {
+            Map.Entry<ContextKey, LibraryContext> entry = iterator.next();
+            if (entry.getKey().executionPlanId() == executionPlanId) {
+                for (TornadoLibraryProvider provider : getProviders()) {
+                    if (provider.libraryName().equals(entry.getKey().libraryName())) {
+                        provider.destroyContext(entry.getValue());
+                        break;
+                    }
+                }
+                iterator.remove();
+            }
+        }
+    }


+    @Override
+    public boolean equals(Object obj) {
+        if (obj instanceof LibraryTask other) {
+            return getId().equals(other.getId());
+        }
+        return false;
+    }
+
+    @Override
+    public int hashCode() {
+        int hash = 7;
+        hash = 71 * hash + Objects.hashCode(getId());
+        hash = 71 * hash + Objects.hashCode(descriptor.getLibraryName());
+        hash = 71 * hash + Objects.hashCode(descriptor.getFunctionName());
+        return hash;
+    }


+        final int n = (int) invocation.getArg(2);
+        final int batch = (int) invocation.getArg(3);


+        final int transa = (int) invocation.getArg(0);
+        final int transb = (int) invocation.getArg(1);
+        final int m = (int) invocation.getArg(2);
+        final int n = (int) invocation.getArg(3);
+        final int k = (int) invocation.getArg(4);


+        final float alpha = (float) invocation.getArg(5);
+        final int lda = (int) invocation.getArg(7);
+        final int ldb = (int) invocation.getArg(9);
+        final float beta = (float) invocation.getArg(10);
+        final int ldc = (int) invocation.getArg(12);


+    target_link_libraries(tornado-cudnn ${JNI_LIB_DIRS} "-lcudnn -lcudart")
+endif()
+
+message(STATUS "TornadoVM cuFFT: include=${CUDA_INCLUDE_DIR}; cudnn=${CUDNN_LIB}")


+    </parent>
+    <artifactId>tornado-cudnn</artifactId>
+    <name>tornado-cudnn</name>
+    <description>TornadoVM cuDNN bindings: NVIDIA cuFFT library tasks for TornadoVM task graphs</description>


github-actions · 2026-07-03T15:25:25Z

⚠️ Could not automatically backport this PR to jdk25.

Cherry-picking commit 4c67d0b64b865870b82045f1cc92ae64a09b66e8 onto jdk25 produced conflicts, so no jdk25 PR was created. Please backport these changes manually.

github-actions · 2026-07-03T15:26:05Z

⚠️ Could not automatically backport this PR to jdk25.

Cherry-picking commit 4c67d0b64b865870b82045f1cc92ae64a09b66e8 onto jdk25 produced conflicts, so no jdk25 PR was created. Please backport these changes manually.

Replace non-ASCII characters (em dashes, middle dots) in javadoc, add trailing periods to arg-list javadoc, wrap an over-long line, and remove ParenPad whitespace in Map.of( // dispatch registries. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

github-actions · 2026-07-03T15:27:31Z

⚠️ Could not automatically backport this PR to jdk25.

Cherry-picking commit 4c67d0b64b865870b82045f1cc92ae64a09b66e8 onto jdk25 produced conflicts, so no jdk25 PR was created. Please backport these changes manually.

github-actions · 2026-07-03T16:10:22Z

⚠️ Could not automatically backport this PR to jdk25.

Cherry-picking commit e4220c2fb05a1b8cb4e1cdfd407359c9e06ba92a onto jdk25 produced conflicts, so no jdk25 PR was created. Please backport these changes manually.

Copilot

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 2 comments.

+    /** (input, output, n, batch). */
+    private static void execC2C(CuFftContext context, LibraryInvocation invocation, int direction) {
+        final int n = (int) invocation.getArg(2);
+        final int batch = (int) invocation.getArg(3);
+



-    private static long getOrCreatePlan(CuFftContext context, PlanKind kind, int dim1, int dim2) {
-        String planKey = kind + ":" + dim1 + ":" + dim2;
+    /** (input, output, n, batch). */


Copilot AI review requested due to automatic review settings July 2, 2026 20:11

github-project-automation Bot added this to JDK PR Backport Tracking Project Jul 2, 2026

Copilot started reviewing on behalf of mikepapadim July 2, 2026 20:11 View session

Copilot AI reviewed Jul 2, 2026

View reviewed changes

stratika removed this from JDK PR Backport Tracking Project Jul 3, 2026

mikepapadim changed the title ~~[hybrid] Hybrid API: native library tasks with cuddn provider~~ [hybrid] Hybrid API: native library tasks with cuDNN provider Jul 3, 2026

mikepapadim mentioned this pull request Jul 3, 2026

[hybrid] Hybrid API: native library tasks with cuFFT provider #887

Merged

github-project-automation Bot added this to JDK PR Backport Tracking Project Jul 3, 2026

stratika removed this from JDK PR Backport Tracking Project Jul 3, 2026

mikepapadim mentioned this pull request Jul 3, 2026

[hybrid] CUDA-C backend: delegate @Reduce to CUB warp reductions #889

Open

mikepapadim marked this pull request as ready for review July 3, 2026 09:30

stratika changed the base branch from master to develop July 3, 2026 10:26

stratika marked this pull request as draft July 3, 2026 15:25

stratika marked this pull request as ready for review July 3, 2026 15:25

stratika closed this Jul 3, 2026

stratika reopened this Jul 3, 2026

stratika added enhancement New feature or request CUDA labels Jul 3, 2026

stratika added this to JDK PR Backport Tracking Project Jul 3, 2026

stratika moved this to Needs backport in JDK PR Backport Tracking Project Jul 3, 2026

Merge branch 'develop' into hybrid-cudnn

316d289

Copilot AI review requested due to automatic review settings July 3, 2026 16:10

Copilot started reviewing on behalf of mikepapadim July 3, 2026 16:10 View session

Copilot AI reviewed Jul 3, 2026

View reviewed changes

mikepapadim merged commit 3cb4c0a into beehive-lab:develop Jul 3, 2026
3 checks passed

yrq0208 moved this from Needs backport to Backport PR Open in JDK PR Backport Tracking Project Jul 6, 2026

stratika mentioned this pull request Jul 6, 2026

Backport Hybrid API to jdk25 #898

Merged

10 tasks

stratika moved this from Backport PR Open to Backporting in JDK PR Backport Tracking Project Jul 6, 2026

stratika moved this from Backporting to Merged to JDK25 in JDK PR Backport Tracking Project Jul 6, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[hybrid] Hybrid API: native library tasks with cuDNN provider#888

[hybrid] Hybrid API: native library tasks with cuDNN provider#888
mikepapadim merged 2 commits into
beehive-lab:developfrom
mikepapadim:hybrid-cudnn

mikepapadim commented Jul 2, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

github-actions Bot commented Jul 3, 2026

Uh oh!

github-actions Bot commented Jul 3, 2026

Uh oh!

github-actions Bot commented Jul 3, 2026

Uh oh!

github-actions Bot commented Jul 3, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

		final int n = (int) invocation.getArg(2);
		final int batch = (int) invocation.getArg(3);

Uh oh!

Conversation

mikepapadim commented Jul 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

What's included

Tests

Benchmarks (RTX 4090, cuDNN 9.23, cross-validated)

Docs

Backend/s tested

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

github-actions Bot commented Jul 3, 2026

Uh oh!

github-actions Bot commented Jul 3, 2026

Uh oh!

github-actions Bot commented Jul 3, 2026

Uh oh!

github-actions Bot commented Jul 3, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

mikepapadim commented Jul 2, 2026 •

edited

Loading